FMRP has a cell-type-specific role in CA1 pyramidal neurons to regulate autism-related transcripts and circadian memory

Summary of read alignment for each replicate of FMRP cTag CA1 CLIP including Fmr1-cTag^Camk2a-Cre and Fmr1-cTag negative control CLIP. The total number of reads obtained after demultiplexing and initial collapse of exact PCR duplicates is given (‘input reads for alignment’) as well as the number of these reads which uniquely mapped to the genome (‘Uniquely mapped to genome’) and transcriptome (‘Mapped within transcriptome and collapsed for PCR duplicates’). Finally, the number of reads which mapped to the positive strand of the CA1 transcriptome (as defined by CA1 TRAP rpkm > 1) is shown (‘Mapped within CA1 TRAP defined transcriptome for CLIP score determination’).

FMRP cTag CLIP data including counts per transcript, CLIP scores, FMRP target classifications and CA1 or granule cell-enriched binding. CA1 CLIP. CLIP tags that mapped within the coding region of each transcript were counted for each CLIP replicate and the final read count calculated by subtracting the number of CLIP reads in the Cre negative control from the number of CLIP reads from the Camk2a-Cre sample (CDS counts). These counts were converted to RPKM based on the length of the coding region. After normalization to abundance using TRAP RPKM per transcript as a proxy, the CLIP score for the transcript was calculated. CA1 Targets. List of transcripts classified as stringent, high binding or low-binding FMRP Targets in CA1 neurons. Stringent targets have a CLIP score > 2 in all three replicates, high binding targets have a mean CLIP score > 1 and low binding targets have a mean CLIP score between 0 and 1. Granule Cell CLIP. FMRP cTag CLIP was performed from cerebellar granule cells using Neurod1-Cre. Table of transcript information and CLIP scores is reproduced from Van Driesche et al. (2019). Granule Cell Targets. List of transcripts classified as stringent, high binding or low-binding FMRP Targets in cerebellar granule cells. Stringent targets have a CLIP score > 2 in all three replicates, high binding targets have a mean CLIP score > 1 and low binding targets have a mean CLIP score between 0 and 1. Enriched CA1. Transcripts with significantly higher binding in CA1 compared to cerebellar granule cells. Differential CLIP scores were determined using Limma. Log₂ fold change and p-values are reported. Enriched Granule Cell. Transcripts with significantly higher binding in cerebellar granule cells compared to CA1 neurons. Differential CLIP scores were determined using Limma. Log₂ fold change and p-values are reported.

Results from Fmr1 KO vs WT TRAP. Raw counts per gene and statistics from DESeq2 analysis of differential expression between WT and KO are shown. Sawicka CA1 Fmr1 KO TRAP DESeq2. CA1 TRAP data from RiboTag^Camk2a-Cre P28-P31 mice presented in this study. Ceolin CA1 Fmr1 KO TRAP DESeq2. CA1 TRAP data from RiboTag^Wfs1-CreERT22–6 month old mice published by Ceolin et al. (2017). Downloaded from GSE94559. Granule Fmr1 KO TRAP DESeq2. Cerebellar granule cell TRAP data from RiboTag^Neurod1-Cre6–8 week mice presented in this study.

CLIP normalized to FACS RNA-Seq. CLIP scores determined from linear model using FACS RNA-Seq transcript RPKM. This table is equivalent to Supplementary file 2 CA1 CLIP but using RPKM per transcript from RNA-Seq of sorted CA1 pyramidal cells to normalize for abundance instead of TRAP. Counts and negative binomial. FMRP targets determined using a count-based method in which the FMRP CLIP read counts driven by mRNA abundance are modelled as following a negative binomial distribution. Reads from CLIP and TRAP replicates that mapped within the coding region of each transcript were counted. For CLIP the final read count was calculated by subtracting the number of CLIP reads in the Cre negative control (‘Neg CLIP CDS reads’) from the number of CLIP reads from the Camk2a-Cre sample (‘Camk2a CLIP CDS reads’). The mean TRAP counts normalized by library size across all three replicates are shown (‘mean normalized TRAP counts’). CLIP scores were derived from a linear regression model as the difference between the observed and fitted values. An estimated dispersion parameter for each CLIP count is given, along with individual p-values per replicate, combined p-value and FDR per transcript.