FMRP promotes RNA localization to neuronal projections through interactions between its RGG domain and G-quadruplex RNA sequences

  1. Raeann Goering
  2. Laura I Hudish
  3. Bryan B Guzman
  4. Nisha Raj
  5. Gary J Bassell
  6. Holger A Russ
  7. Daniel Dominguez
  8. J Matthew Taliaferro  Is a corresponding author
  1. Department of Biochemistry and Molecular Genetics, University of Colorado Anschutz Medical Campus, United States
  2. Barbara Davis Center for Diabetes, University of Colorado Anschutz Medical Campus, United States
  3. Department of Pharmacology, University of North Carolina at Chapel Hill, United States
  4. Departments of Cell Biology and Neurology, Emory University School of Medicine, Georgia
  5. RNA Bioscience Initiative, University of Colorado Anschutz Medical Campus, United States
7 figures, 1 table and 2 additional files

Figures

Figure 1 with 11 supplements
Identification of FMRP localization targets.

(A) Schematic of soma/neurite fractionation. Cells are plated on porous membranes. Neurites grow down through the pores, and cells are then mechanically fractionated by scraping. (B) Western blot of …

Figure 1—figure supplement 1
Location of guide RNA directed against Fmr1.

Sanger sequencing tracks of wildtype and knockout CAD cells are shown. In all alleles of the knockout cells, a single basepair deletion was created, leading to the creation of a premature stop codon …

Figure 1—figure supplement 2
RNA expression levels of Fmr1 in the wildtype and FMRP null cells.
Figure 1—figure supplement 3
PCA analysis of gene expression values from the soma and neurite compartments of wildtype and FMRP null cells.
Figure 1—figure supplement 4
Correlation and clustering analysis of gene expression values from the soma and neurite compartments of wildtype and FMRP null cells.
Figure 1—figure supplement 5
LR values for all genes (gray), ribosomal protein genes (red), and genes that are part of the electron transport chain (blue).

We have observed that RNAs encoding ribosomal protein genes and components of the electron transport chain are consistently neurite-enriched and therefore use them as markers to assess whether the …

Figure 1—figure supplement 6
Clustering of LR values from 65 neuronal subcellular transcriptomic experiments.
Figure 1—figure supplement 7
Standard deviation of LR values for each gene across all samples.

The control distribution was constructed by shuffling gene/LR relationships. The P value was calculated using a Wilcoxon rank-sum test.

Figure 1—figure supplement 8
LR values for genes encoding ribosomal proteins for all samples.

Red samples are those generated in the current study.

Figure 1—figure supplement 9
LR values for genes encoding members of the electron transport chain for all samples.

Red samples are those generated in the current study.

Figure 1—figure supplement 10
LR values for genes in wildtype and FMRP null cells.

Those that met an FDR cutoff of 0.01, as determined by Xtail, are shown in red.

Figure 1—figure supplement 11
Gene ontology enrichments for genes with reduced neurite localization in FMRP knockout CAD cells.
Figure 2 with 10 supplements
Identification of transcript features associated with the regulation of localization by FMRP.

(A) Fraction of FMRP localization targets (red) and nontargets (gray) that were previously identified as directly bound by FMRP in cells. FMRP-bound RNAs were defined as the intersection between …

Figure 2—figure supplement 1
Overlap between FMRP functional localization targets and FMRP CLIP-seq targets as identified in Ascano et al.

P values were calculated using a binomial test.

Figure 2—figure supplement 2
Change in RNA localization in FMRP knockout cells compared to wildtype for FMRP CLIP-seq targets (red) and non-targets (gray) as identified by Ascano et al.

P values were calculated using a Wilcoxon rank-sum test.

Figure 2—figure supplement 3
Overlap between FMRP functional localization targets and FMRP CLIP-seq targets as identified in Darnell et al.

P values were calculated using a binomial test.

Figure 2—figure supplement 4
Change in RNA localization in FMRP knockout cells compared to wildtype for FMRP CLIP-seq targets (red) and non-targets (gray) as identified by Darnell et al.

P values were calculated using a Wilcoxon rank-sum test.

Figure 2—figure supplement 5
Overlap between FMRP functional localization targets and FMRP CLIP-seq targets as identified in Maurin et al.

P values were calculated using a binomial test.

Figure 2—figure supplement 6
Change in RNA localization in FMRP knockout cells compared to wildtype for FMRP CLIP-seq targets (red) and non-targets (gray) as identified by Maurin et al.

P values were calculated using a Wilcoxon rank-sum test.

Figure 2—figure supplement 7
Experimentally-defined G-quadruplex sequences contain more guanosine residues than control sequences drawn from the same genes.
Figure 2—figure supplement 8
Minimum free energy structures calculated using RNAfold show the G-quadruplex structures are significantly more stable than control sequences drawn from the same genes.
Figure 2—figure supplement 9
Experimentally defined G-quadruplex sequences contain many more regular expression-defined G-quadruplex sequences than controls.
Figure 2—figure supplement 10
Experimentally defined G-quadruplex sequences contain many more RNAfold-defined quadruplexed guanosine residues than control sequences.
Figure 3 with 6 supplements
RNA mislocalization in FXS neurons.

(A) FMRP protein expression in motor neurons differentiated from iPS cells derived from unaffected and FXS patients. (B) FMR1 transcript abundance in motor neurons differentiated from iPS cells …

Figure 3—figure supplement 1
Schematic overview of motor neuron differentiation (top) and representative cell images (bottom).
Figure 3—figure supplement 2
Protein dot blot assaying the efficiency of motor neuron fractionations.

Beta-actin is a marker of both soma (S) and neurite (N) fractions while Histone H3, being restricted to the nucleus, is a marker of soma fractions.

Figure 3—figure supplement 3
LR values for all genes (gray), ribosomal protein genes (red), and genes that are part of the electron transport chain (blue).
Figure 3—figure supplement 4
PCA analysis of gene expression values from the soma and neurite compartments of wildtype and FXS motor neurons.
Figure 3—figure supplement 5
Correlation between FMRP-dependent changes in LR values observed in mouse (FMRP-null CAD - wildtype CAD) and human (FXS motor neurons - unaffected motor neurons) neuronal cells.
Figure 3—figure supplement 6
LR values for genes observed in unaffected and FXS motor neurons.

Genes whose log2 LR value changed by more than 0.25 were designated as either more localized in FXS (gray) or less localized in FXS (red).

Figure 4 with 1 supplement
RNA Bind-n-Seq (RBNS) of FMRP domains.

(A) Schematic representation of RBNS assay and domain architecture of FMRP. (B) Coomassie stain of purified recombinant FMRP fragments used for RBNS experiments. (C) Enrichments for G-quadruplex RNA …

Figure 4—figure supplement 1
Fluorescence polarization values from FMRP RGG/RNA interaction assays.
Figure 5 with 2 supplements
Efficient localization of a reporter transcript requires both a G-quadruplex sequence and the RGG domain of FMRP.

(A) FMRP rescue constructs and smFISH reporter constructs used. smFISH probes, represented by red bars with asterisks, hybridize to the ORF of both reporter constructs. (B) Expression of FMRP rescue …

Figure 5—figure supplement 1
Summary of smFISH results.

Effect sizes and p values are reported.

Figure 5—figure supplement 2
Western blotting of IP samples from the CLIP-seq qPCR experiments.
Figure 6 with 3 supplements
The RGG domain of FMRP is required for the localization of G-quadruplex-containing transcripts transcriptome-wide.

(A) Correlation of LR values for all expressed genes in FMRP rescue samples. (B) G-quadruplex density, as measured by RNAfold, in the 3′ UTRs of transcripts that were more localized in the GFP …

Figure 6—figure supplement 1
LR values for all genes (gray), ribosomal protein genes (red), and genes that are part of the electron transport chain (blue).
Figure 6—figure supplement 2
Correlation of delta LR values (nonfunctional FMRP condition [GFP, RGG, or KO] - functional FMRP condition [WT or FMRP]) between experiments.
Figure 6—figure supplement 3
G-quadruplex density, as measured by RNAfold, in the 3′ UTRs of transcripts that were more localized in the GFP rescue cells (dark gray, left) or FMRP-RGG rescue cells (red, right).

P values were calculated using Wilcoxon rank-sum tests. As in Figure 2D, values represent the number of guanosine residues participating in quadruplex sequences per nt.

Figure 7 with 11 supplements
The translational regulation targets of FMRP are distinct from its RNA localization targets and have no relationship to G-quadruplex sequence density.

(A) Schematic of how the removal of FMRP, which is known to cause ribosome stalls, may affect overall ribosome density on a transcript. (B) Fraction of genes whose ribosome occupancy significantly …

Figure 7—figure supplement 1
Fraction of reads that map to the indicated genomic regions.
Figure 7—figure supplement 2
Fraction of reads of that are the indicated lengths.
Figure 7—figure supplement 3
Fraction of reads that map to the indicated reading frames.
Figure 7—figure supplement 4
Metagene analysis of read densities across transcripts.

Reads from ribosome protected fragments display a characteristic three nucleotide periodicity.

Figure 7—figure supplement 5
Ribosome occupancy values, as calculated by Xtail, in wildtype and FMRP null CAD cells.
Figure 7—figure supplement 6
Gene ontology analysis of the translational regulatory targets of FMRP in CAD cells.
Figure 7—figure supplement 7
Metagene analysis of GC content across transcripts whose ribosome occupancy increases (blue) or decreases (red) in FMRP null CAD cells compared to wildtype cells.
Figure 7—figure supplement 8
Effect sizes and significance of the relationships between transcript features and whether or not a transcript’s ribosome occupancy was increased or decreased by FMRP.
Figure 7—figure supplement 9
Quantification of FMRP trafficking using immunofluorescence.

For each cell, the proportion of the total cellular fluorescence that was present in the neurite was quantified. P values were calculated using a Wilcoxon rank-sum test.

Figure 7—figure supplement 10
G-quadruplex density, as measured by RNAfold, in the 3′ UTRs of transcripts that were more localized in the I304N rescue cells (dark gray, left) or wildtype FMRP rescue cells (red, right).

P values were calculated using Wilcoxon rank-sum tests.

Figure 7—figure supplement 11
Expression of the different HA-tagged FMRP rescue constructs used in the FMRP null CAD line.

The FMRP antibody used was raised against the C-terminus, and therefore does not react with the ΔRGG truncation.

Tables

Key resources table
Reagent type
(species) or resource
DesignationSource or referenceIdentifiersAdditional information
Gene (Mus musculus)Fmr1ENSMUSG00000000838
Sequence-based reagentFmr1 guide RNAUsed to create mouse FMR1-null cells, cloned into pX330
AAATTATCAGCTGGTAATTT
Cell line (Mus musculus)CADSigma08100805-1VL, RRID:CVCL_0199
Cell line (Mus musculus)CAD/loxPKhandelia et al., 2011Contains single integration of loxP cassette
Transfected construct (Mus musculus)Fmr1 cDNADharmaconBC079671.1
AntibodyMouse anti FMR1, monoclonalProteintech665481:5000 dilution for immunoblotting
AntibodyMouse anti HA, monoclonalGenscriptGenScript Cat# A01244, RRID:AB_12893061:2000 for immunofluorescence
Sequence-based reagentsmFISH probes against Firefly luciferaseBioSearchVSMF-1006–5
Commercial assay or kitZymo Quick-RNA Microprep kitZymo ResearchR1050
OtherCell culture inserts for fractionatoinCorning353102

Additional files

Supplementary file 1

Xtail outputs for differential localization or ribosome occupancy of transcripts between two different conditions.

(a) Xtail output for the differential localization of transcripts in wildtype and FMRP null CAD cells. All log2 fold change values are knockout/wildtype. (b) Xtail output for the differential localization of transcripts in unaffected and FXS motor neurons. All log2 fold change values are FXS/unaffected. (c) Xtail output for the differential localization of transcripts in FMRP null CAD cells rescued with either GFP or full length FMRP. (d) Xtail output for the differential localization of transcripts in FMRP null CAD cells rescued with either FMRP-RGG or full length FMRP. (e) Xtail output for the differential localization of transcripts in FMRP null CAD cells rescued with either FMRP-RGG or GFP. (f) Xtail output for the differential ribosome occupancy of genes in wildtype and FMRP null CAD cells. (g) Xtail output for the differential localization of transcripts in FMRP null CAD cells rescued with either GFP or I304N FMRP. (h) Xtail output for the differential localization of transcripts in FMRP null CAD cells rescued with either I304N or wildtype FMRP.

https://cdn.elifesciences.org/articles/52621/elife-52621-supp1-v1.xlsx
Transparent reporting form
https://cdn.elifesciences.org/articles/52621/elife-52621-transrepform-v1.docx

Download links