Polyglycine-mediated aggregation of FAM98B disrupts tRNA processing in GGC repeat disorders

Abstract

Aggregation-prone polyglycine-containing proteins produced from expanded GGC repeats are implicated in an emerging family of neurodegenerative disorders. In this study, we showed that polyglycine itself forms aggregates that incorporate endogenous glycine-rich proteins, including FAM98B, a component of the transfer RNA (tRNA) ligase complex (tRNA-LC) that harbors the most glycine-rich sequence in the human proteome. Through this glycine-rich intrinsically disordered region (IDR), polyglycine sequesters and depletes the tRNA-LC, disrupting tRNA processing. Accordingly, patient tissues revealed aggregate-associated FAM98B depletion and accumulation of aberrant tRNA splicing intermediates. Furthermore, Fam98b depletion in adult mice caused progressive motor coordination deficits and hindbrain pathology. Our data suggest that the FAM98B glycine-rich IDR mechanistically links previously disparate neurodegenerative disorders of protein aggregation and tRNA processing.

Fig. 1.. PolyGly forms aggregates enriched for endogenous glycine-rich proteins.

(A) Schematic of DOX-repressible polyGly expression vectors. (B) Immunofluorescence microscopy of 15xGly or 99xGly HEK293T cells in the presence or absence of DOX; dashed boxes mark magnified regions. Scale bars, 100 μm or 10 μm (inset). (C) Immunofluorescence microscopy of 15xGly or 99xGly HEK293T cells in the absence of DOX. Scale bars, 10 μm. (D) Schematic of polyGly aggregate purification procedure. (E and F) Immunoblotting (E) and protein staining (F) of input, soluble IP, and insoluble IP fractions from polyGly aggregate purification. (G) Enrichment among aggregate-enriched proteins (n=54) compared to the human proteome (n=20,379) of residues within all predicted IDRs (“IDR”), or within predicted IDR subtypes: glycine-rich (“GR”), proline-rich (“PR”), polyampholyte (“PA”), low complexity (“LC”), positive polyelectrolyte (“PPE”), polar (“PO”), or negative polyelectrolyte (“NPE”); inset shows the cumulative distribution functions of GR IDR residues among aggregate-enriched proteins and the human proteome. (H and I) PolyG score profiles at each of the first 16 residues of GAR1 (H) and at each residue position for GAR1 and METAP2 (I); the overall polyG score for each protein is the maximal polyG score at any position in its sequence; predicted GR IDRs are highlighted in gray. (J) Enrichment among aggregate-enriched proteins (n=54) compared to the human proteome (n=20,379) of amino acid homopolymer scores; inset shows the cumulative distribution functions of polyG scores among aggregate-enriched proteins and the human proteome. Experiments in (E) and (F) were performed independently at least three times with consistent results.

Fig. 2.. PolyGly aggregates recruit the tRNA-LC.

(A) GO term enrichment among aggregate-enriched proteins (n=54) compared to the human proteome (n=20,598). (B) AlphaFold-multimer prediction of the structure of the tRNA-LC. (C) DISOPRED3 disorder prediction (top), amino acid composition (middle), and polyG score profile (bottom) for FAM98B. (D) Distribution of polyG scores in the human proteome. (E) Immunoblotting of input and insoluble IP fractions from polyGly aggregate purification for tRNA-LC components. (F) Immunofluorescence microscopy for tRNA-LC components in 15xGly or 99xGly HEK293T cells in the absence of DOX. Scale bars, 10 μm. Experiment in (E) was performed independently twice with consistent results.

Fig. 3.. PolyGly post-translationally depletes the tRNA-LC.

(A) Quantitative proteomics of soluble lysates from clonal 15xGly and 99xGly HEK293T cells in the presence or absence of DOX (n=3 clones per condition). (B) GO term enrichment among up- (n=150) and down-regulated (n=61) proteins in (A) compared to all quantitated proteins (n=8,807). (C) Immunoblotting of soluble and insoluble fractions from 15xGly or 99xGly HEK293T cells in the absence of DOX for the indicated duration. Results are representative of 4 independent clones. (D) Immunoblotting of soluble lysates from HEK293T cells transfected with 15xGly or 99xGly and treated with cycloheximide (CHX) for the indicated duration. (E) Immunoblotting of soluble lysates from HEK293T cells transfected with 15xGly or 99xGly and extracted in the indicated protein extraction buffer. (F) Immunoblotting of soluble lysates from HEK293T cells transfected with 15xGly or 99xGly and treated with the indicated compound. Experiments in (C) to (F) were performed independently at least twice with consistent results.

Fig. 4.. Pathological and native polyGly interact to deplete the tRNA-LC.

(A) Schematic of 15xGly, 99xGly, 99xGly^STOP, 99xAla, and 99xArg expression vectors. (B) Immunoblotting of soluble and insoluble lysates from HEK293T cells transfected with 15xGly, 99xGly, 99xGly^STOP, 99xAla, or 99xArg. (C) Schematic of FAM98B^FL and FAM98B^ΔC. (D and E) Immunoblotting of soluble and insoluble lysates (D) and immunofluorescence microscopy (E) of HEK293T cells stably expressing FAM98B^FL or FAM98B^ΔC and transfected with 15xGly or 99xGly. Scale bars, 10 μm. (F) Immunoblotting of soluble and insoluble lysates from HEK293T cells stably expressing DD-FAM98B^ΔC and transfected with 15xGly or 99xGly in the absence or presence of Shield-1; E and DD denote endogenous FAM98B and DD-FAM98B^ΔC, respectively. Experiments in (B), (D), and (F) were performed independently at least twice with consistent results.

Fig. 5.. PolyGly disrupts the biogenesis of intron-containing tRNAs.

(A) Schematic of spliced tRNA biogenesis. (B) Northern blotting with Tyr-GTA 5′ and 3′ exon probes in HEK293T cells transfected with the indicated siRNA. (C) Quantitation of tRNA band intensities by northern blotting with the indicated probe in HEK293T cells transfected with the indicated siRNA; intensities are normalized to U6 and expressed relative to control siRNA. (D) Northern blotting with Tyr-GTA 5′ and 3′ exon probes in 15xGly and 99xGly HEK293T cells in the presence or absence of DOX. (E) Northern blotting with the indicated probe in 15xGly and 99xGly HEK293T cells in the presence or absence of DOX. (F) Quantitation of tRNA band intensities by northern blotting with the indicated probe in 15xGly and 99xGly HEK293T cells in the presence or absence of DOX; intensities are normalized to U6 and expressed relative to the +DOX intensity for the respective clonal line. Error bars, SEM (n=3 clones per condition). (G) Northern blotting with Tyr-GTA-2–1 leader and trailer probes in parental HEK293T cells (WT) and clonal FAM98B knockout cells (KO 1 and KO 2) transfected with empty vector (EV), FAM98B^FL (FL), or FAM98B^ΔC (ΔC). (H) Northern blotting with Tyr-GTA-2–1 leader and trailer probes in HEK293T cells stably expressing DD-FAM98B^ΔC and transfected with 15xGly or 99xGly in the absence or presence of Shield-1. Experiment in (H) was performed independently twice with consistent results.

Fig. 6.. Intranuclear inclusions recruit and deplete the tRNA-LC in GGC repeat diseases.

(A and B) Immunofluorescence microscopy for tRNA-LC components in FXTAS cerebellum (A) and NIID skin (B). Scale bars, 5 μm. (C and D) Quantitation of soluble FAM98B fluorescence intensity in inclusion-lacking or -containing nuclei in FXTAS pons (C) and NIID skin (D); the mean intensity across inclusion-containing nuclei normalized to that of inclusion-lacking nuclei is shown in red. (E) Quantitation of soluble FAM98B fluorescence intensity in inclusion-lacking or -containing nuclei in the indicated FXTAS brain region; the mean intensity across inclusion-containing nuclei normalized to that of inclusion-lacking nuclei is shown in red. (F) Quantitation of soluble intensities of the indicated tRNA-LC components in inclusion-lacking or -containing nuclei in FXTAS cerebellum; the mean intensity across inclusion-containing nuclei normalized to that of inclusion-lacking nuclei is shown in red.

Fig. 7.. tRNA processing is disrupted in FXTAS.

(A) Normalized coverage of 5’ (left) and 3’ (right) pre-tRNA-derived fragments from spliced tRNAs, averaged across 4 unaffected control and 4 FXTAS cerebellum samples. (B) Normalized counts of 3’ ends of 5’ fragments (left) and 5’ ends of 3’ fragments (right) of pre-tRNA-derived fragments from spliced tRNAs. Error bars, SEM (n=4 individuals per group). (C) Distribution of Wald statistics for differential abundance of pre-tRNA-derived 5’ halves (left) and 3’ halves (right) from non-spliced or spliced tRNAs. (D) Northern blotting with the indicated probe in unaffected control and FXTAS cerebellum. (E) Quantitation of tRNA fragment band intensities by northern blotting with the indicated probe in unaffected control and FXTAS cerebellum; intensities are normalized to 5.8S rRNA and expressed relative to the mean of unaffected controls. Error bars, SEM (n=4 individuals per group). Normalization in (A) to (C) was performed based on miRNA counts.

Fig. 8.. Fam98b depletion is sufficient to cause neuropathology in vivo.

(A) Schematic of experimental approach to deplete Fam98b from adult mice using CRISPR/Cas9. (B) Northern blotting with the indicated probe in brain RNA of mice injected with AAV encoding a non-targeting control sgRNA or one of two sgRNAs targeting Fam98b. (C) Quantitation of tRNA band intensities by northern blotting with the indicated probe in brain RNA of mice injected with AAV encoding the indicated sgRNA; intensities are normalized to U6 and expressed relative to the mean of sgLacZ-injected mice (n=4 mice per group). (D) Time before falling from a rotating rod for mice injected with AAV encoding the indicated sgRNA (n=9 sgLacZ, 6 sgFam98b_g1, and 6 sgFam98b_g2). (E) Immunofluorescence microscopy for Gfap in brainstems of mice injected with AAV encoding the indicated sgRNA. Scale bars, 50 μm. (F) Quantitation of Gfap⁺ area within brainstems of mice injected with AAV encoding the indicated sgRNA (n=4 mice per group). (G) Model of tRNA splicing dysfunction in GGC repeat disease: polyGly-containing intranuclear inclusions sequester and deplete the tRNA-LC via the FAM98B glycine-rich IDR, resulting in impaired tRNA splicing and accumulation of unligated tRNA exons. Experiments in (D) to (F) were performed independently twice with consistent results.