METTL17 is an Fe-S cluster checkpoint for mitochondrial translation

Abstract

Friedreich's ataxia (FA) is a debilitating, multisystemic disease caused by the depletion of frataxin (FXN), a mitochondrial iron-sulfur (Fe-S) cluster biogenesis factor. To understand the cellular pathogenesis of FA, we performed quantitative proteomics in FXN-deficient human cells. Nearly every annotated Fe-S cluster-containing protein was depleted, indicating that as a rule, cluster binding confers stability to Fe-S proteins. We also observed depletion of a small mitoribosomal assembly factor METTL17 and evidence of impaired mitochondrial translation. Using comparative sequence analysis, mutagenesis, biochemistry, and cryoelectron microscopy, we show that METTL17 binds to the mitoribosomal small subunit during late assembly and harbors a previously unrecognized [Fe₄S₄]²⁺ cluster required for its stability. METTL17 overexpression rescued the mitochondrial translation and bioenergetic defects, but not the cellular growth, of FXN-depleted cells. These findings suggest that METTL17 acts as an Fe-S cluster checkpoint, promoting translation of Fe-S cluster-rich oxidative phosphorylation (OXPHOS) proteins only when Fe-S cofactors are replete.

Keywords: FA; Fe-S cluster; Friedreich’s ataxia; METTL17; frataxin; mitochondria; mitoribosome.

Fig 1:. Proteomic analysis of FXN edited cells reveals a marked depletion of known Fe-S cluster containing proteins and reduction of small mitoribosome subunits.

A. Quantitative whole cell proteomic analysis was carried out on K562 cells edited with control or FXN targeted guides in duplicate, depleting for this allosteric regulator of Fe-S cluster biosynthesis. B. Waterfall plot of protein fold change in FXN/Control edited cells, highlighting FXN and validated human Fe-S cluster containing proteins. C. OXPHOS proteins are organized by complex with blue indicating proteins that are depleted in FXN edited cells. Genes are ordered alphabetically within complex using complex-specific prefixes (NDUF, SDH, UQCR, COX, ATP5) (e.g. A2 in CI refers to NDUFA2 whereas A in CII refers to SDHA). D. Waterfall plot of protein fold change in FXN/Control edited cells, highlighting proteins in the small and large mitoribosome subunit, as well as the small subunit assembly factor, METTL17. E. Cumulative distribution of Fe-S containing proteins and proteins of the small mitoribosome subunit versus all identified proteins in FXN/Control edited cells. ****=p < 0.0001. Two sample Kolmogorov-Smirnov test. See also Fig S1 and Table S2.

Fig 2:. Mitochondrial translation is attenuated in the absence of FXN.

A. Left-Mitochondrial translation, as assessed by autoradiography after ³⁵S-methionine/cysteine labeling, of cells expressing sgRNAs targeting FXN, NDUFS1 or FBXL5. All cells were treated with 200μg/mL emetine, and control cells in the last lane were also treated with 50μg/μL chloramphenicol. Right-Quantification of the band intensities in control vs. FXN edited cells. B. Schematic overview of the genome-wide CRISPR genetic interaction screens carried out in K562 cells. Cells were either infected with guides against FXN or a control locus before introduction of the library. Following expansion, cells were sequenced to assess the relative abundance of guides in the FXN depleted vs. control background. C. GO (top) and MitoCarta 3.0 (bottom) enrichment analysis of genetic interactors identified in for FXN depleted cells. In parentheses are the number of genes enriched in the screen for each term. D. Scatterplot of Z scores showing knockouts growth in sgCtrl vs. sgFXN backgrounds. The positive control (IRP2) and the mitochondrial ribosome assembly genes (METTL17 and MPV17L2) are highlighted. See also Fig S2 and Table S3.

Fig 3:. METTL17 is depleted in the absence of FXN and is essential for robust mitochondrial translation.

A. Immunoblot for FXN, METTL17 and the loading control actin in K562 cells edited with control, FXN, NDUFS1 and FBXL5 guides. B. qPCR for METTL17 expression levels in sgCtrl and sgFXN cells (n=8 for all samples) C. Cells edited for control, FXN, METTL17 and CDK5RAP1 genes were grown for 24h in galactose media, and viability was assessed for each background (n=4 for all samples) . D. Immunoblot for FXN, METTL17, CDK5RAP1, select OXPHOS subunits and the loading control tubulin in cells edited with control, FXN, METTL17 and CDK5RAP1 guides. E. Correlation analysis of gene dependencies sourced from DepMap. Presented is the gene network that correlates with METTL17 deletion using FIREWORKS. Solid and dashed lines represent primary and secondary correlations, respectively. F. Mitochondrial translation, as assessed by autoradiography after ³⁵S-methionine/cysteine labeling, of cells expressing sgRNAs targeting METTL17 or CDK5RAP1. All cells were treated with 200μg/mL emetine, and control cells in the last lane were also treated with 50μg/μL chloramphenicol. G. qPCR analysis of 12S levels in cells edited with control, FXN, METTL17 or CDK5RAP1 guides (n=8 for all samples). H. Immunoblot for METTL17 and the loading control actin in cells edited with control or FXN guides. Following editing, cells were grown in 21% or 1% oxygen. All bar plots show mean ± SD. **=p < 0.01, ****=p < 0.0001. Two-way ANOVA with Bonferroni’s post-test. See also Fig S3.

Fig 4:. METTL17 harbors conserved motifs for Fe-S binding that are crucial for its functionality.

A. Multiple sequence alignment for METTL17 homologues, highlighting two motifs associated with Fe-S cluster binding; 4 cysteine metal binding pocket (red) and a LYR handoff motif (blue). B. Immunoblot from whole cell and mitoprep extracts of cells expressing GFP, METTL17-FLAG, CYS^Mut-FLAG or LYR^Mut-FLAG constructs. C. Control or METTL17 edited cells expressing GFP, METTL17-FLAG, CYS^Mut-FLAG or LYR^Mut-FLAG constructs were grown for 24h in galactose, following which their viability was assessed (n=4–6). D. Immunoblots examining OXPHOS subunits or the loading control TUBULIN in Control or METTL17 edited cells expressing GFP, METTL17-FLAG, CYS^Mut-FLAG or LYR^Mut-FLAG constructs. E. Mitochondrial translation, as assessed by autoradiography after ³⁵S-methionine/cysteine labeling, in Control or METTL17 edited cells expressing GFP, METTL17-FLAG, CYS^Mut-FLAG or LYR^Mut-FLAG constructs. All cells were treated with 200μg/mL emetine, and control cells in the last lane were also treated with 50μg/μL chloramphenicol. F. qPCR analysis of 12S levels in Control or METTL17 edited cells expressing GFP, METTL17-FLAG, CYS^Mut-FLAG or LYR^Mut-FLAG constructs (n=3 for all samples). G. Formaldehyde-linked RNA immunoprecipitation of the 12S to GFP, METTL17-FLAG, CYS^Mut-FLAG or LYR^Mut-FLAG proteins. Results were normalized to input construct and 12S levels (n=2 for all samples). All bar plots (except panel G) show mean ± SD. ****=p < 0.0001. Two-way ANOVA with Bonferroni’s post-test. See also Fig S4.

Fig 5:. Human METTL17 expressed and purified from E. coli contains an Fe-S cluster.

A. Gel filtration chromatography and SDS-PAGE analysis demonstrate that the purified METTL17 construct runs as a monomer near its predicted molecular weight of 50 kD. B. Iron content of purified METTL17and CYS^Mut as determined by bicinchoninic acid assay and inductively coupled plasma mass spectrometry. The CYS^Mut is a METTL17 variant in which the four cysteines predicted to coordinate the cluster are mutated to serine (C333S, C339S, C347S, and C404S). C. The UV-Vis absorption spectra of METTL17 exhibits a broad band around 420 nm, consistent with the presence of an [Fe₄S₄]²⁺ or an [Fe₃S₄]⁺ cluster; the latter is ruled out by EPR spectroscopy as described in the text. This band is lost in the spectrum of CYS^Mut. For clarity, spectra were normalized to the intensity at 280 nm.

Fig 6:. Cryo-EM structure of the yeast SSU-METTL17 complex and key functional elements.

A. Cryo-EM density maps obtained for the SSU with METTL17 (left) or mtIF3 (right). B. Overall view of METTL17 on the SSU, and close-up views. Top close up shows the position of METTL17 (C-terminal domain, CTD sky blue; N-terminal domain, NTD blue) between the rRNA (yellow) of the head and body, while the C-terminal extension (CTE) occupies the mRNA path. Bottom close up shows the coordination of 4Fe-4S cluster by four cysteines, including Cys513 from the CTD, and related structural elements with their cryo-EM densities: flipped base A1100, a cis-proline, arginine that is within salt bridge distance, and a conserved histidine that can be involved in a transfer and ligation to the Fe-S unit. C. Conformational changes within the rRNA region h30–34 that is involved in METTL17 binding. Superimposed models of the SSU-METTL17 (yellow) with unbound state (grey). Sticks represent the rRNA residues responsible for the METTL17 interaction. D. Superposition of SSU-METTL17 with SSU-mtIF3 showing clashes of METTL17 (blue) with mtIF3 (orange surface representation). See also Table S1, Fig S5 and S6.

Fig 7:. Overexpression of METTL17 restores the mitochondrial bioenergetics, but not growth, of FXN depleted human cells.

A. Control or FXN edited cells expressing GFP, METTL17-FLAG, CYS^Mut-FLAG or LYR^Mut-FLAG constructs were grown for 24h in glucose (left) or galactose (right), following which their viability was assessed (n=4–6). B. Immunoblots examining OXPHOS subunits or the loading control TUBULIN in Control or FXN edited cells expressing GFP, METTL17-FLAG, CYS^Mut-FLAG or LYR^Mut-FLAG constructs. C. Oxygen consumption rate (OCR) of Control or FXN edited cells expressing GFP or METTL17-FLAG. Cells were sequentially treated with oligomycin, Bam15 and piericidin+antimycin. D. Population doubling over 72h of Control or FXN edited cells expressing GFP, METTL17-FLAG, CYS^Mut-FLAG or LYR^Mut-FLAG constructs (n=4 for all samples). E. FXN activates Fe-S cluster formation, which can be utilized to support (i) mitochondrial bioenergetics via formation of the electron transport chain (ETC) or (ii) cell growth and division. METTL17 is a key Fe-S cluster bearing modulator of mitochondrial bioenergetics, and its absence in FXN depleted cells accounts for much of the mito. bioenergetic defects observed in these cells. All bar plots show mean ± SD. **=p < 0.01, ****=p < 0.0001. Two-way ANOVA with Bonferroni’s post-test. See also Fig S7.