The Kelch 3 motif on gigaxonin mediates the interaction with NUDCD3 and regulates vimentin filament morphology

Abstract

Gigaxonin is an intermediate filament (IF)-interacting partner belonging to the Kelch-like (KLHL) protein family. Gigaxonin is encoded by the KLHL16 gene, which is mutated in Giant Axonal Neuropathy (GAN). The lack of functional gigaxonin in GAN patient cells impairs IF proteostasis by affecting IF protein degradation and transport. This leads to focal abnormal accumulations of IFs and compromised cellular function, with neurons being most severely impacted. We hypothesized that gigaxonin forms molecular interactions via specific sequence motifs to regulate IF proteostasis. The goal of this study was to examine how distinct Kelch motifs on gigaxonin regulate IF protein degradation and filament morphology. We analyzed vimentin IFs in HEK293 cells overexpressing wild type (WT) gigaxonin, or gigaxonin lacking each of the six individual Kelch motifs, K1-K6. All six gigaxonin deletion mutants (ΔK1-ΔK6) promoted the degradation of soluble vimentin. Compared to WT-gigaxonin, ΔK3-gigaxonin exhibited increased soluble vimentin degradation and increased presence of thick bundles of vimentin IFs. The ΔK4 mutant showed similar, but milder phenotypes compared to ΔK3. Using mass spectrometry proteomics we found that, relative to WT gigaxonin, ΔK3 gigaxonin had increased associations with ubiquitination-associated and mitochondrial proteins but lost the association with the NudC domain-containing protein 3 (NUDCD3), a molecular chaperone enriched in the nervous system. AlphaFold modeling revealed loss of gigaxonin-NUDCD3 binding with ΔK3 and altered binding with ΔK4. Collectively, our cell biological data show the induction of an abnormal GAN-like IF phenotype in cells expressing ΔK3- and, to a lesser extent, ΔK4-gigaxonin, while our proteomic profiling links the loss of gigaxonin-NUDCD3 interactions with defective IF proteostasis.

Keywords: Chaperone; Cytoskeleton; Gigaxonin; Intermediate filaments; Kelch domain; NUDCD3; Neurodegenerative disease.

Figure 1.. Pathogenicity of GAN missense variants across the gigaxonin protein domains.

A. Missense mutations are the most common GAN variants; based on a genetic analysis of the GAN natural history study cohort in Bharucha-Goebel et al., Brain 2021. B. AlphaMissense scores on known disease-causing GAN variants from the natural history study, as represented by box-and-whisker plot (n=22; median score=0.9850). C. Missense GAN variants from ClinVar plotted by their respective cDNA location (x-axis) and AlphaMissense pathogenicity score (y-axis). Black dots represent known disease-causing GAN variants; from Bharucha-Goebel et al., Brain 2021 and Lescouzeres & Bomont, Front Physiol 2020. Blue and red dots represent variants where blue is predicted benign and red is predicted pathogenic. Yellow boxes represent the gigaxonin functional domains BTB, BACK, and Kelch, as defined in UniProt (ID: Q9H2C0). D. Frequency distribution plot of AlphaMissense pathogenicity scores of gigaxonin variants, according to the protein domain affected: BTB, green; BACK, blue; KELCH; magenta.

Figure 2.. Effect of WT and Kelch deletion mutant gigaxonin over-expression on soluble vimentin levels.

A. Schematic representation of the gigaxonin domains and the location of the turbo-GFP tag at the C-terminus. B. Immunoblot for wild type GFP-tagged and untagged gigaxonin and the effect of their overexpression on TritonX-soluble vimentin levels in HEK293 cells; actin serves as a loading control. C. Quantification of the vimentin immunoblots from panel B. *p<0.05; one-way ANOVA. D. AlphaFold model of gigaxonin showing the predicted beta propeller structure and location of each Kelch motif that compose the gigaxonin Kelch domain. E. Immunoblots of GFP-gigaxonin and endogenous gigaxonin in cells expressing WT or Kelch deletion mutants. Two exposures of the same membrane are shown to visualize the relative levels of the endogenous vs. the over-expressed GFP-tagged form. F. Immunoblot of TritonX-soluble vimentin in HEK293 cells transfected with WT and Kelch deletion mutants of gigaxonin. G. Quantitative vimentin ELISA of soluble vimentin. **p<0.01; one-way ANOVA compared to WT-Gig group.

Figure 3.. Vimentin filament morphology in HEK293 cells expressing WT and Kelch deletion mutants of gigaxonin.

Immunofluorescence analysis of GFP-gigaxonin (green), vimentin (magenta), and DAPI (cyan) in HEK293 cells transfected with GFP-tagged WT gigaxonin (Gig) and Kelch motif deletion mutants (ΔK1-K6); Lipofectamine-only condition serves as transfection control. Scale bar=20μm. Boxed areas are enlarged to show the filament phenotypes in transfected cells.

Figure 4.. Quantification of vimentin filament bundling.

A. Representative examples of normal vimentin filaments (left; Lipofectamine-only control condition) and bundled vimentin filaments (right; ΔK3-Gig transfection) in HEK293 cells. Shown side-by-side are the raw images and local thickness - analyzed images with colors representing pixel intensity in grey value. Scale bars=10 μm. B. Histograms of local thickness intensity counts for each condition. *p<0.05; **p<0.01; ***p<0.001; two-way ANOVA. C. Box-and-whisker plot of combined counts for pixels with intensity value >40 for each condition. **p<0.01; one-way ANOVA.

Figure 5.. Effect of gigaxonin Kelch 3 and Kelch 4 deletion on gigaxonin protein interactions.

A. Volcano plot of protein interactors of ΔK3-gigaxonin that were significantly changed from WT-gigaxonin, based on mass spectrometry analysis on GFP-Gig pulldowns. Cyan dots represent increased interactions; magenta dot represents decreased interaction. B. Heat maps of the two major categories of differentially interacting gigaxonin partners in the absence of the Kelch 3 motif. Left map shows ubiquitin ligases/deubiquitinases and right map shows mitochondrial proteins (selected out of 73 statistically significant proteins detected in the ΔK3-Gig pulldown relative to WT-Gig; full list in Supplemental Table 2). C. Heat map of gigaxonin interactors that were significantly changed in K3 and K4 compared to WT-Gigaxonin. D. Immunoblot analysis of NUDCD3 and GFP-Gig in GFP pulldowns on cell lysates from Lipofectamine control, GFP empty vector and GFP-gigaxonin conditions (WT, ΔK3 and ΔK4). Input lanes show total levels of NUDCD3 and actin (loading control) in the lysates. E. Immunofluorescence analysis of vimentin and NUDCD3 localization in two GAN patient fibroblasts (B15-100.1; top and B16-02; bottom). Boxed areas denote magnified images. Arrowheads point to co-localized signal between vimentin aggregates and NUDCD3 and arrows point to lack of co-localization between vimentin aggregates and NUDCD3. Scale bar = 20μm.

Figure 6.. AlphaFold 3 predicted model of gigaxonin interacting with NUDCD3.

Gigaxonin is shown in green with the exception of the Kelch domain that is colored by motif: Kelch 1 (aa274-326, yellow), Kelch 2 (aa327-374, blue), Kelch 3 (aa376-421, red), Kelch 4 (aa422-468, black), Kelch 5 (470-522, purple), Kelch 6 (528-574, orange). Upon rotating the model 90 degrees, the classic beta-propeller organization of the Kelch repeat motifs becomes visible, showing the NUDCD3 (in cyan) interactions with specific Kelch motifs. B. Zoomed in image of AlphaFold 3 predicted model of gigaxonin (colored by Kelch motif as described above, interacting with NUDCD3 (cyan). Amino acids in the Kelch domain predicted to be within 3Å of NUDCD3 are localized to Kelch 4-6 and are shown as green sticks. C. AlphaFold 3 predicted model of gigaxonin ΔKelch 3 (aa376-421) (green) interacting with NUDCD3 (cyan), with the remaining Kelch motifs colored as in panel A. With the removal of Kelch 3, NUDCD3 is no longer predicted to interact with the Kelch domain. D. AlphaFold 3 predicted model of gigaxonin ΔKelch 4 (aa422-468) (green) interacting with NUDCD3 (cyan), with the remaining Kelch motifs colored as in panel A. Without Kelch 4, NUDCD3 is predicted to adopt an entirely different conformation to interact with gigaxonin. This includes NUDCD3 interacting more with Kelch 1 (yellow), especially through its N-terminal (green sticks), which was not shown in the wild-type predicted model.