The FXR1 network acts as a signaling scaffold for actomyosin remodeling

Abstract

It is currently not known whether mRNAs fulfill structural roles in the cytoplasm. Here, we report the fragile X-related protein 1 (FXR1) network, an mRNA-protein (mRNP) network present throughout the cytoplasm, formed by FXR1-mediated packaging of exceptionally long mRNAs. These mRNAs serve as an underlying condensate scaffold and concentrate FXR1 molecules. The FXR1 network contains multiple protein binding sites and functions as a signaling scaffold for interacting proteins. We show that it is necessary for RhoA signaling-induced actomyosin reorganization to provide spatial proximity between kinases and their substrates. Point mutations in FXR1, found in its homolog FMR1, where they cause fragile X syndrome, disrupt the network. FXR1 network disruption prevents actomyosin remodeling-an essential and ubiquitous process for the regulation of cell shape, migration, and synaptic function. Our findings uncover a structural role for cytoplasmic mRNA and show how the FXR1 RNA-binding protein as part of the FXR1 network acts as an organizer of signaling reactions.

Keywords: RNA-binding protein; biomolecular condensates; cytoplasmic organization; fragile X syndrome; messenger ribonucleoprotein network; non-canonical roles of mRNA; signal transduction; signaling scaffold; spatial proximity; structural role of mRNA.

Figure 1.. FXR1 assembles with its bound mRNAs into a cytoplasmic mRNP network.

A. Immunofluorescence staining of endogenous FXR1 protein in HeLa cells. The dotted line indicates the nucleus. Right panel is a zoomed-in image of the region in the yellow box. All cells contain the network, and a representative confocal image is shown. All scale bars in Figure 1 are 5 μm. B. Live cell confocal imaging of HEK293T cells with endogenous monomeric NG-tagged FXR1 protein. All cells contain the network and a representative image is shown. C. Size exclusion chromatography of cells from (B), immunoblotted for FXR1. CLUH was used as loading control. mNG-FXR1 and FXR1 have the same elution pattern. D. IUpred2A score of human FXR1. A score greater than 0.5 indicates an IDR. Schematics of GFP-fusion constructs. The numbers denote amino acids. E. Live cell confocal imaging of HeLa cells transfected with the FXR1 constructs from (D) after knockdown of endogenous FXR1. The GFP fluorescence pattern shown for each construct was observed in all cells expressing the respective FXR1 constructs. Representative images are shown. See Fig. S4C for quantification. F. Confocal imaging of HeLa cells transfected with GFP-FXR1-N2 after digitonin permeabilization in the presence or absence of RNase A treatment for 30 minutes. Representative images from at least three independent experiments are shown, where 21 cells were examined. G. FRAP analysis of GFP-FXR1 full-length (FL), -N1, and -N2 expressed in HeLa cells. Shown is a normalized FRAP curve as mean ± std from at least 11 cells each. MF, mobile fraction. See Videos S7-9 for representative fluorescence recovery. Mann-Whitney test, N1 vs. FL, ***, P<10⁻²¹; N2 vs. FL, ***, P<10⁻¹⁶⁵.

Figure 2.. FXR1 dimerization through the CC domains promotes mRNA binding and nucleates the FXR1 network.

A. Amino acid boundaries of FXR1 protein domains. Domains capable of binding to RNA or protein are indicated. B. Schematic of FXR1 CC mutant constructs and their resulting FXR1 network assembly states. Red star symbols represent single point mutations. CC1mut is N202P, CC2mut is V361P. See Fig. S4A-E for details. C. Live cell confocal imaging of HeLa cells transfected with GFP-FXR1 constructs from (B) after knockdown of endogenous FXR1, shown as in Fig. 1A. Representative images from at least three independent experiments are shown where 38 cells were examined. See Fig. S4C for quantification. Scale bar, 5 μm. D. GFP co-IP of endogenous FXR1 after ectopic expression of GFP-FXR1-WT or GFP-FXR1-CC2mut in HeLa cells. Actin is shown as loading control. 1% of input was loaded. E. Oligo(dT) pulldown, performed without cross-linking, of mRNA-bound FXR1 in FXR1/FXR2/FMR1 triple knockout (KO) U2OS cells after ectopic expression of GFP or GFP-FXR1 constructs from (B). The endogenously expressed RNA-binding protein HuR was used as positive and loading control for oligo(dT)-bound proteins. 2.5% and 5% of input were loaded in the left and right panels, respectively.

Figure 3.. The FXS mutations I304N and G266E disrupt the FMR1 and FXR1 networks.

A. Amino acid boundaries of FMR1 protein domains and schematics of FMR1 constructs. B. Live cell confocal imaging of HeLa cells transfected with GFP-FMR1 constructs from (A), shown as in Fig. 1A. All cells with WT-FMR1 contain the network and most cells with mutant FMR1 show network disruption (see Fig. S4C for quantification). Representative images are shown. Scale bar, 5 μm. C. Oligo(dT) pulldown, performed without cross-linking, of mRNA-bound FMR1 in FXR1/FXR2/FMR1 triple KO U2OS cells after ectopic expression of GFP or GFP-FMR1 constructs from (A). The endogenous RNA-binding protein HuR was used as positive and loading control for oligo(dT)-bound proteins. 1% of input was loaded. D. Live cell confocal imaging of HeLa cells transfected with GFP-FXR1-WT or FXS mutant constructs after knockdown of endogenous FXR1, shown as in (B). The FXS mutations G266E and I304N are located at the same amino acid positions in FXR1 and FMR1. The network is disrupted in all cells (see Fig. S4C for quantification). Representative images are shown. E. As in (C), but oligo(dT) pulldown was performed after ectopic expression of GFP-FXR1-WT, -G266E, or -I304N. F. Sanger sequencing results of heterozygous and homozygous N202S CC1-disrupting point mutations in endogenous FXR1 in A549 clonal cells generated using base editing. G. Oligo(dT) pulldown of mRNA-bound FXR1 in A549 clonal cells from (F). Endogenous HuR was used as positive and loading control for oligo(dT)-bound proteins. 1% of input was loaded. H. Quantification of FXR1-bound mRNAs from (G) shown as mean ± std obtained from three independent experiments. One-way ANOVA, *** P<0.001. I. Live cell confocal imaging of A549 clonal cells from (F) after knockin of monomeric GFP into the endogenous FXR1 locus. Scale bar, 5 μm. J. Sanger sequencing results of heterozygous KH1 domain point mutation G266E in endogenous FXR1 in A549 clonal cells generated using prime editing. K. Oligo(dT) pulldown of mRNA-bound FXR1 in A549 clonal cells from (J). Endogenous HuR was used as positive and loading control for oligo(dT)-bound proteins. 0.2% of input was loaded. L. Quantification of FXR1-bound mRNAs from (K) shown as mean ± std obtained from three independent experiments. One-way ANOVA, *, P<0.05, **, P<0.01. M. Schematic summarizing the results from (F) to (L).

Figure 4.. The FXR1 network is required for RhoA signaling-induced actomyosin reorganization.

A. All mRNAs expressed in HeLa cells are grouped based on their FXR1 binding pattern. mRNAs not bound by FXR1 (N=6574), bound by FXR1 but network-independent (N=1104), bound by FXR1 and network-dependent (N=1223). Boxes represent median, 25^th and 75^th percentiles, error bars represent 5-95% confidence intervals. Mann-Whitney test, ***, P<10⁻⁵³. B. As in (A), but mRNA length is shown. Mann-Whitney test, ***, P<10⁻¹⁴. C. As in (A), but AU-content of mRNAs is shown. Mann-Whitney test, ***, P<10⁻⁵⁴. D. Gene ontology analysis for FXR1 network-dependent mRNA targets. Shown are the top functional gene classes and their Bonferroni-corrected P values. E. Schematic of RhoA signaling pathway-induced actomyosin remodeling. The critical signaling event for actomyosin dynamics is RLC phosphorylation of NM II. Protein symbols with black outlines are FXR1 mRNA targets. ELC, essential light chain. P, phosphorylated residue. F. Phalloidin staining of filamentous actin in A549 cells expressing the indicated shRNAs after serum starvation and stimulation with thrombin for 30 minutes. DAPI staining visualizes the nucleus. Representative images are shown. Scale bar, 40 μm. G. Quantification of the experiment in (F) shown as mean ± std obtained from at least three independent experiments. For each experiment and each sample at least 150 cells were counted, except for the ROCK2 knockdown experiment, where 34 cells were counted. One-way ANOVA, ****, P<0.0001. H. As in (F), but A549 clonal cells with heterozygous N202S mutations in endogenous FXR1 were used. Shown are representative images. I. Quantification of the experiment in (H) shown as mean ± std obtained from at least three independent experiments. For each experiment and each sample at least 28 cells were counted. One-way ANOVA, ****, P<0.0001. J. As in (F), but A549 clonal cells with heterozygous G266E mutations in endogenous FXR1 were used. Shown are representative images. K. Quantification of the experiment in (J) shown as mean ± std obtained from three independent experiments. For each experiment and each sample, at least 70 cells were counted. One-way ANOVA, ****, P<0.0001. L. Fraction of migrated A549 cells for the indicated samples is shown as mean ± std from at least three independent experiments. One-way ANOVA, ****, P<0.0001, **, P<0.01, *, P<0.05, n.s., not significant.

Figure 5.. Phosphorylation of RLC by ROCK2 kinase is FXR1 network dependent.

A. Tandem Mass Tag quantitative proteomics analysis of HeLa cells after control or FXR1 knockdown. Proteins whose abundance was significantly affected by FXR1 knockdown are colored red (N=6), whereas proteins not significantly affected are colored in blue (N=7061). B. Western blot of the indicated endogenous proteins in A549 cells grown in the indicated conditions. Ctrl, expressing control shRNA, KD, expressing FXR1 shRNA1. TCP1 was used as loading control. C. Quantification of phospho-RLC level from (B) shown as mean ± std obtained from at least three independent experiments. One-way ANOVA, **, P< 0.01. n.s., not significant. D. Western blot of the indicated proteins in serum-starved and thrombin-stimulated parental A549 and clonal cell lines containing WT FXR1 or a heterozygous N202S mutation in endogenous FXR1. α-Tubulin was used as loading control. E. Quantification of phospho-RLC level from (D) shown as mean ± std obtained from three clonal cell lines each. One-way ANOVA, **, P<0.01. F. Western blot of the indicated proteins in serum-starved and thrombin-stimulated parental A549 and clonal cell lines containing WT FXR1 or a heterozygous G266E mutation in endogenous FXR1. α-Tubulin was used as loading control. G. Quantification of phospho-RLC level from (F) shown as mean ± std obtained from two clonal cell lines each. H. Schematic of the proximity ligation assay (PLA), which generates a positive signal if the distance between two endogenous proteins is smaller than 40 nm. I. PLA performed in serum-starved thrombin-stimulated A549 cells, indicating that FXR1 is required for proximity between ROCK2 and RLC, but not for proximity between ROCK2 and MYPT1. As negative control, the RLC antibody alone was used. DAPI staining visualizes the nucleus. Representative images are shown. Scale bar, 20 μm. J. Quantification of the experiment in (I), shown as mean ± std of three independent experiments. For each experiment and each sample at least 39 cells were counted. One-way ANOVA, ****, P<0.0001.

Figure 6.. FXR1 network-dependent protein interactors contain CC, Tudor, and RGG domains.

A. SILAC mass spectrometry analysis of HeLa cells. Shown is log2 fold change (FC) of protein counts of CC2mut/WT samples. Reduced interaction in CC2 mutant samples indicates that the interaction with FXR1 is network dependent. The top network dependent FXR1 interactors are indicated. For full list, see Table S3. B. Validation of the SILAC proteomics results using GFP co-IP of the indicated endogenous proteins followed by western blot in the presence or absence of RNase A. GFP-FXR1 constructs were ectopically expressed in HeLa cells depleted of endogenous FXR1. 0.5% input was loaded. C. As in (B), but GFP co-IP of endogenous FXR1 by ectopically expressed interactors. The red star symbol marks an unspecific band. 1% input was loaded. D. Protein domains of the top FXR1 network-dependent interactors. Shown are CC, Tudor, RG/RGG, and R-rich domains in color. E. Fold enrichment of indicated protein domains in the 20% of proteins from (A) with the most negative FC. Shown is the observed-over-expected frequency. Chi-square test, **, P=0.002, ***, P<0.0001. Chi-square test for Tudor domains cannot be performed as the numbers are too small. See Table S3.

Figure 7.. The presence of CC, Tudor, or RGG domains is sufficient for binding to FXR1.

A. Protein domains of NM II (MYH9), RLC (MYL9), and ROCK2. Highlighted are CC and R-rich domains. B. Amino acid boundaries of ROCK2 protein domains and schematics of ROCK2 constructs. The numbers indicate amino acids. C. GFP co-IP, followed by western blot of endogenous FXR1 after ectopic expression of GFP-ROCK2-C or GFP-ROCK2-C-ΔCC (from B) in HeLa cells. 1% input was loaded. D. Schematic of GFP-GAPDH fusion constructs. The following domains were fused to GAPDH: CC2 domain of ROCK2, RGG domain of TOP3B, Tudor domain of TDRD3, R-rich region of TDRD3, and both the Tudor and R-rich regions of TDRD3. See Fig. S7E for their amino acid sequences. E. GFP co-IP, followed by western blot of endogenous FXR1 after ectopic expression of GFP-GAPDH fusion constructs from (D) in HeLa cells. A representative experiment is shown. F. Quantification of (E). Shown is FXR1 enrichment normalized to sample C2 (shown in magenta) as mean ± std obtained from at least three independent experiments. G. Model of the FXR1 network and its function as a scaffold for signaling reactions by establishing spatial proximity between kinases and their substrates. P, phosphorylated residue. PPI, protein-protein interaction. See text for details.