Inhibition of RNA splicing triggers CHMP7 nuclear entry, impacting TDP-43 function and leading to the onset of ALS cellular phenotypes

Abstract

Amyotrophic lateral sclerosis (ALS) is linked to the reduction of certain nucleoporins in neurons. Increased nuclear localization of charged multivesicular body protein 7 (CHMP7), a protein involved in nuclear pore surveillance, has been identified as a key factor damaging nuclear pores and disrupting transport. Using CRISPR-based microRaft, followed by gRNA identification (CRaft-ID), we discovered 55 RNA-binding proteins (RBPs) that influence CHMP7 localization, including SmD1, a survival of motor neuron (SMN) complex component. Immunoprecipitation-mass spectrometry (IP-MS) and enhanced crosslinking and immunoprecipitation (CLIP) analyses revealed CHMP7's interactions with SmD1, small nuclear RNAs, and splicing factor mRNAs in motor neurons (MNs). ALS induced pluripotent stem cell (iPSC)-MNs show reduced SmD1 expression, and inhibiting SmD1/SMN complex increased CHMP7 nuclear localization. Crucially, overexpressing SmD1 in ALS iPSC-MNs restored CHMP7's cytoplasmic localization and corrected STMN2 splicing. Our findings suggest that early ALS pathogenesis is driven by SMN complex dysregulation.

Keywords: ALS; CHMP7; CRISPR screen; RNA splicing; SMN complex; SmD1; TDP-43; amyotrophic lateral sclerosis; neurodegeneration.

Figure 1:. Image-based genome-wide CRISPR screen technologies identifies modulators of CHMP7 nuclear localization.

A. Schematic of the Craft-ID experiment. A CRISPR–Cas9 gRNA library was generated from an array of sgRNA oligonucleotides cloned into the lentiCRISPR v2 backbone. HeLa cells expressing GFP-CHMP7 were infected at low multiplicity of infection (MOI .15) and cultured in bulk for 7 d after selection, allowing lethal guides to drop out of the pool. A bulk infection of cells with a gRNA library targeting over 1,000 annotated proteins (>12,000 sgRNAs) was performed, followed by single cell plating on 12 microRaft arrays to screen genetic knockout clones for mislocalization of CHMP7. Positive candidates, where GFP is in the nucleus and co-localizing with nuclear staining by DAPI, were selected. To sequence the sgRNA associated with CHMP7 mislocalization to the nucleus, we isolated target colonies adhered to microRafts from the array. A motorized microneedle, fitted over the microscope objective, was actuated to pierce the PDMS microarray substrate and dislodge individual magnetic microRafts from the array. Released microRafts and their cargo were collected with a magnetic wand into a strip tube containing a lysis buffer for a targeted two-step PCR with in-line barcodes, followed by high-throughput sequencing. B. Bar chart showing the total number of rafts picked (84), sequenced (73 with successfully obtained PCR products), identified (55 using CRaft-ID), and proteins confirmed by siRNA depletion (23). The pie chart represents the distribution of sgRNAs identified from the 84 rafts. C. Gene Ontology analysis of the 55 candidate genes identified in the screen compared to the background of genes targeted in the CRISPR screen.

Figure 2:. RNA processing plays a pivotal role in governing the cellular localization of CHMP7.

A. Representative images of proteins XPO7, DHX8, and SmD1 depleted by siRNAs in HeLa cells. Immunofluorescent analysis of DAPI (blue), Phalloidin (Actin filaments, green) and CHMP7 (magenta). Scale bars, 50 μm. B-D. Western blot analysis of XPO7, SmD1 and DHX8 in cells treated with targeting siRNAs, alongside NTC controls. E. Quantification of image intensity of cytoplasmic (C) and nuclear (N) CHMP7 in HeLa cells when XPO7, DHX8 and SmD1 are targeted by siRNA, compared to NTC, represented by the y-axis label as N/C ratio. Data are presented as mean ± SD of three independent experiments (n=3 wells, ~800 cells total). Statistical significance analyzed by Student’s t-test, ****P < 0.0001.

Figure 3:. CHMP7 interacts with SMN complex proteins responsible for snRNP assembly in iPSC-motor neurons.

A. Schematic of IP-MS workflow to identify CHMP7 protein-protein interactions with or without RNase treatment. B-C. Volcano plots showing proteins significantly enriched in untreated or RNase-treated CHMP7 IPs compared to IgG IPs. The x-axis shows log2 fold-changes between CHMP7 IP to IgG, and the y-axis shows -log10 p-values (unpaired Student’s t-test). Proteins with p-values < 0.01 and FC > 2 are labeled as interactors. D-E. Gene ontology terms that were enriched for the protein interactors in untreated or RNase-treated samples. The -log10 (p-value) depicting the statistical significance of enrichment is plotted for each GO term. Background of all the proteins in IPs was implemented in the GO term analysis. A list of enriched interactors can be found in Table S1. F. Network analysis was performed for CHMP7 interactors across all GO term pathways using the MCODE algorithm to identify densely connected protein neighborhoods in the network. The biological roles of each component are annotated. G. CHMP7 RNA-mediated interactors associate with snRNP assembly proteins (SMD1–3), GEM, and nuclear pore complex proteins. Nodes are colored based on k-means clustering and edge confidence; high (0.700), highest (0.900), medium (0.400) from String.

Figure 4:. CHMP7 binds RNA, specifically RNA processing targets, in iPSC-motor neurons

A. Hydra analysis of CHMP7 protein, with annotations for low complexity regions, and high Δscores represent regions that occlusion analysis indicated to be predicted RNA binding domains (RBDs). B. Pie chart of genomic features represented in enriched windows in CHMP7 eCLIP data from day 28 iPSC-MNs. C. Bar plot of gene features of CHMP7 eCLIP binding enrichment (fold change) normalized to size-matched input control in day 28 iPSC-MNs. SKIPPER analysis was used to compute these enrichments (Table S2). D. Comparison of CHMP7 eCLIP to ENCODE3 eCLIP data, clustered by binding preferences to genic features. E. Statistically significantly enriched Gene Ontology terms for CHMP7 RNA targets. F. Heatmap indicating relative information for snoRNA and snRNA elements with CHMP7 binding. G. Example of significantly enriched CHMP7 eCLIP binding site on snRNA gene with size-matched input control reads on the same scale

Figure 5:. Downregulation of RNA processing upon CHMP7 nuclear accumulation in HeLa cells.

A. Immunofluorescence staining of GFP-CHMP7, Helix 6 (GFP-CHMP7^NES2–) or Helix 5 and Helix 6 (GFP-CHMP7^{NES1-&NES2–}) with nuclear DAPI signal in HeLa cells. Scale bars, 50 μm. B. Volcano plot comparing differentially expressed transcripts from RNA-seq analysis of GFP-GFP-CHMP7^NES2 vs. GFP-CHMP7 transient expression, with p-value threshold of 0.05 and FC of 2 as cutoffs. C. Gene Ontology analysis of downregulated and upregulated transcripts from volcano plot in B (Table S3). D. Distribution of classes of differentially changing alternative splicing events (FDR≤0.05+|ΔΨ|<0.1) comparing GFP-CHMP7^NES2– to GFP-CHMP7 expression.

Figure 6:. Alterations in CHMP7 protein-RNA landscape in ALS iPSC-motor neurons.

A-C. Number of enriched CHMP7 binding windows across control, C9orf72 and sALS lines. D. Skipper t-SNE analysis of CHMP7 enriched binding in genic features across control, C9orf72, sALS, with ENCODE eCLIP data colored by target preferences in the background. E. Mean relative information around introns for CHMP7 control, C9orf72, and sALS in lines. F. Mean relative information around 5’UTR for CHMP7 in control and sALS lines.

Figure 7:. Overexpression of SmD1 can sufficiently rescue CHMP7 cytoplasmic levels in sALS iPSC-motor neurons.

A. Immunofluorescence of day 28 iPSC-MNs with either NTC and si-SmD1, stained with CHMP7 and TDP-43. Showing CHMP7 nuclear localization in SmD1 KD and reduction in nuclear TDP-43. Scale bars, 10 μm. B. Quantification of nuclear (N) to cytoplasmic (C) CHMP7 intensity in motor neurons (n=3, ~150 cells total). C. Quantification of nuclear TDP-43 levels with si-SmD1 compared to NTC (n=3, ~50 cells total). D-E. Overexpression of control plasmid and SmD1 open-reading frame at day 36 in sALS lines showing mRNA levels for STMN2 and truncated STMN2 (n=4). F. Quantification of nuclear CHMP7 in neurons overexpressing SmD1 (n=3, ~35 cells total). G. Immunofluorescence staining of neurons overexpressing SmD1 in sALS at day 36, stained with CHMP7, MAP2, and DAPI. Scale bars, 10 μm. H. mRNA levels of SmD1 in 12 sALS lines and 8 C9orf72, normalized to GAPDH and control lines at day 32. The data are presented as the mean ± SD. Significance was assessed using Student’s t-test (*p < 0.05, ***p < 0.001, ****p < 0.0001).

Figure 8:. Perturbation of SMN-Complex-Dependent snRNP assembly modulates CHMP7 nuclear influx localization and alters NPC proteins.

A. Live cell imaging of GFP-CHMP7 and TDP-43-mCherry at 0- and 120-minutes post SMN inhibitor treatment. Heatmap of fluorescent intensity ratio of TDP-43 at 0 min and at 120 min. Scale bars, 20 μm. B. X-axis indicates minutes (min); 0 min is prior to SMN inhibitor treatment. Nuclear GFP intensity, cells quantified, is ~12 used and averaged from three independent experiments. Error bars show ± SD. C. Immunofluorescence analysis of CHMP7 (Green) and DAPI with DMSO or SMN inhibitor in HeLa after 2- hours (n = 3). Scale bars, 100 μm. D. Proposed mechanism of action. Created with BioRender.com