Identification of substrates of the human mitochondrial ClpXP protease and its implications for Perrault syndrome

Abstract

The ClpXP ATP-dependent unfoldase-protease complex plays important roles in maintaining protein homeostasis in the mitochondria. Pathogenic variants of CLPP have been identified in patients with Perrault syndrome type 3 (PRLTS3), a rare recessive mitochondrial disease characterized by sensorineural hearing loss, ovarian dysfunction, and neurological abnormalities. However, it remains unclear how these ClpP variants cause this disease. Here we identify three cellular substrates of ClpXP using proteomic analysis of PRLTS3 patient-derived fibroblasts, with follow up studies using induced pluripotent stem cells (iPSCs) and iPSC-derived neural progenitor cells, alongside in vitro biochemical assays. These substrates are: CHCHD2, a bi-organellar regulator of oxidative phosphorylation and mitochondrial stress responses; ALAS1, which catalyzes the rate-limiting step in heme biosynthesis; and TFAM, an essential mitochondrial transcription factor. Efficient degradation of these substrates by ClpXP requires the adaptor protein PDIP38. We propose that dysregulated cellular levels of these substrates might contribute to the etiology of PRLTS3.

The ClpXP ATP-dependent unfoldase-protease complex plays important roles in maintaining protein homeostasis in the mitochondria. Here, the authors identify three substrates of ClpXP that are dysregulated in PRLTS3 patients and show that degradation of these substrates requires the adaptor PDIP38.

Introduction

Regulation of mitochondrial protein homeostasis is essential for cellular metabolism and is often dysregulated in disease¹. Under physiological conditions, mitochondrial chaperones and proteases maintain proteostasis in this organelle by folding nascent proteins or by the timely removal of aggregated and misfolded proteins². Caseinolytic protease P (ClpP) is one of two major proteases localized to the mitochondrial matrix of eukaryotes that regulate proteins involved in metabolic pathways and the electron transport chain (ETC)^3–5. Structurally, ClpP forms a cylindrical tetradecameric oligomer, comprised of fourteen protomers that assemble into two stacked heptameric rings, which enclose a proteolytic chamber with fourteen Ser-His-Asp catalytic triads^6,7. In humans, substrate specificity for ClpP-mediated proteolysis requires association with ClpX, an AAA+ ATP-dependent unfoldase that belongs to the Hsp100 family of molecular chaperones⁶. ClpX has an N-terminal zinc binding domain (ZBD), a C-terminal AAA+ ATP-binding and hydrolysis domain, and an additional E-domain that is inserted within the AAA+ domain^8–10. Functionally, ClpX docks as a hexamer onto the ClpP cylinder using its LGF loops that bind to surface-exposed hydrophobic pockets between neighboring ClpP protomers to form an active ClpXP complex^11,12. Once complexed, ClpX recognizes substrates, unfolds them in an ATP-dependent manner, and subsequently threads them into the ClpP proteolytic chamber for degradation¹³. In the absence of ClpX, ClpP is unable to degrade fully folded proteins, though it can degrade small peptides¹². Several substrates have been proposed for ClpXP, including proteins involved in the ETC, the tricarboxylic acid (TCA) cycle, fatty acid and amino acid metabolism, and mitochondrial transcription and translation⁴. However, many of these putative substrates have been identified through proteomic screens in ClpP knockout models and to date have yet to be directly validated⁴.

Rare biallelic variants within CLPP (the gene encoding ClpP) are associated with Perrault syndrome type 3 (PRLTS3), a rare autosomal recessive mitochondrial disorder that is characterized by bilateral sensorineural hearing loss, ovarian dysfunction or azoospermia in female or male patients, respectively, and, in severe cases, neurological abnormalities such as ataxia, epilepsy, intellectual disability, and/or neuropathology¹⁴. Biallelic variants in at least nine separate genes, including CLPP, TWNK, ERAL1, HARS2, LARS2, DAP3, MRPL49, GPPS1, and PEX6 have collectively been implicated in Perrault syndrome, where dysregulation of mitochondrial protein synthesis and homeostasis, alongside peroxisomal dysfunction, appears to play a common etiological role^14–16. In relation to ClpP, over twenty unique disease-related CLPP variants have been documented from PRLTS3 patients, most containing single nucleotide variants^14,17,18. These variants have been documented throughout the ClpP protein sequence, and while structure-based predictions exist, very few of these Perrault-related CLPP variants have been characterized in cells or in vitro. It is currently unclear how CLPP variants, and the resultant ClpP dysfunction, contribute to Perrault syndrome, and identification of ClpP substrates relevant to Perrault syndrome remains to be elucidated.

Herein, we employ PRLTS3 patient-derived fibroblasts, induced pluripotent stem cells (iPSCs) and iPSC-derived neural progenitor cells (NPCs), CLPP^−/− knockout cells, and in vitro purified protein models to identify three bona fide ClpXP protein substrates. These include CHCHD2, ALAS1, and TFAM, and we demonstrate that their ClpXP-mediated degradation is significantly enhanced by the adaptor protein PDIP38. Additionally, we have characterized multiple PRLTS3-related CLPP variants biochemically in vitro and performed an in-depth analysis of two distinct variants, Cys144Arg (c.403 T > C) and Thr145Pro (c.433 A > C), to identify mechanisms for PRLTS3-related ClpP dysfunction. These results identify molecular and cellular outcomes of CLPP variants that may be critical to the etiology of Perrault syndrome.

Results

Perrault-related ClpP variants show differences in steady-state ClpP protein levels

To better understand the relationship between ClpP and Perrault syndrome, we generated expression vectors for fourteen C-terminally 3×FLAG-tagged ClpP containing individual missense mutations reported from patients with Perrault syndrome, as well as for wild-type (WT) ClpP (Fig. 1a). When superimposed onto the crystal structure of recombinant human ClpP (PDB: 1TG6)¹¹, several of these mutations appear to cluster around the ClpX-interacting hydrophobic pocket (e.g. C144R and T145P), while others are localized within or near the mitochondrial targeting sequence (MTS; e.g. P34S and L58R) or active site (e.g. Y229D). While prior in vivo modeling has shown that Clpp^−/− knockout mice exhibit Perrault-like phenotypes^19,20, the steady-state expression levels of these Perrault-related ClpP proteins is currently unknown. Therefore, we transiently expressed FLAG-tagged ClpP variants in human HEK293 T-REx CLPP^−/− knockout cells and assessed steady-state ClpP protein levels by anti-FLAG immunoblotting. Overall, multiple variants displayed reduced steady-state protein levels when compared to WT-ClpP, including L58R-, R78P-, C86Y-, P142L-, T145P-, and G162S-ClpP (Fig. 1b). This was not due to differences in mRNA transcript levels as measured by RT-qPCR (Supplementary Fig. 1). Interestingly, other Perrault-related ClpP variants, such as P34S-, T135S-, C144R-, C147S-, I208M-, A210T-, E221K-, and Y229D-ClpP appeared to express similarly to WT-ClpP (Fig. 1b), suggesting an alternative mechanism unrelated to ClpP expression by which these variants may cause ClpP dysfunction. Thus, to elucidate these differing mechanisms, we selected C144R- and T145P-ClpP as two neighboring candidate ClpP variants with, respectively, unaffected and affected steady-state protein levels for further cellular studies.

Fig. 1. Steady-state protein levels of variant ClpP implicated in Perrault syndrome type 3.

a Schematic representation of Perrault-related variants in human ClpP as reported in the literature. Mutated residues are highlighted in red, while the catalytic triad (S153, H178, D227) is denoted with cyan on the monomeric protein structure of human WT-ClpP (yellow; PDB: 1TG6)¹¹. b Steady-state protein levels of 3×FLAG-tagged ClpP variants transiently expressed in HEK293 T-REx CLPP^−/− cells and normalized to tubulin. Control cells were transfected with empty vector (*p ≤ 0.05, ****p ≤ 0.0001; one-way ANOVA with Bonferroni’s post-hoc test, mean ± SD, n = 4 biological replicates). From left to right, p-values are: 0.013, 0.015, <0.0001, <0.0001, <0.0001, <0.0001. Source data are provided in the Source Data file.

Proteomic analysis of T145P-ClpP (c.433 A > C) patient-derived skin fibroblasts identifies candidate ClpXP substrates

Given the crucial role of ClpP-mediated proteolysis in maintaining mitochondrial homeostasis, we next sought to identify ClpXP substrates that are dysregulated in Perrault syndrome. Thus, we performed mass spectrometry analysis on patient-derived primary skin fibroblasts harboring a homozygous CLPP^T145P/T145P (c.433 A > C) variant²¹. As controls, mass spectrometry was performed on age-matched wild-type (CLPP^WT/WT) skin fibroblasts from an unrelated healthy donor alongside skin fibroblasts from a heterozygous (CLPP^T145P/WT) sibling. A total of 5893 unique proteins were detected, and comparisons were segmented into CLPP^T145P/T145P vs CLPP^WT/WT, CLPP^T145P/T145P vs CLPP^T145P/WT, and CLPP^T145P/WT vs CLPP^WT/WT, and analyzed for mitochondrial (Fig. 2a) and total proteins (Supplementary Fig. 2a). Multiple proteins that have been previously implicated in Perrault syndrome displayed significant changes in their abundance in CLPP^T145P/T145P cells, including MRPL49²², HSD17B4²³, MRPL50²⁴, HARS2²⁵, GGPS1¹⁶, and PEX6¹⁶ (Fig. 2a; relative fold-changes reported in Supplementary Data 1). Interestingly, a decrease in abundance was also observed for proteins involved in the ETC and oxidative phosphorylation, such as multiple NDUF subunits (NDUFB3, NDUFV3, NDUFS4, NDUFA9, NDUFA12)²⁶ and COX7A1²⁶, alongside the metabolic proteins ALDH1B1²⁷, AASS²⁸, and SQOR²⁹ (Fig. 2a, Supplementary Data 1), suggesting gross changes in mitochondrial activity and metabolism in Perrault-related cells.

Fig. 2. Proteomic analysis of PRLTS3 patient-derived T145P-ClpP (c.433 A > C) skin fibroblasts.

a Volcano plot depicting fold-change comparisons in the mitochondrial proteome profiles of Perrault-patient CLPP^T145P/T145P versus heterozygous (CLPP^T145P/WT) or control age-matched wild-type (CLPP^WT/WT) skin fibroblasts. Proteins of interest are highlighted with a grey rectangle. For significance, the -log₁₀(p-value) cutoff was set to 1.3 (p = 0.05; student’s two-tailed t-test with Benjamini-Hochberg correction, n = 4 biological replicates). b, c Gene set enrichment analysis (GSEA) was performed on data from (a), and gene ontologies (GOs) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were plotted according to q-values (color gradient; cut-off of ≤ 0.15). Set size is indicated by the circle size, shown on the right, for each GO term and KEGG pathway. d Steady-state protein levels in Perrault-patient CLPP^T145P/T145P, heterozygous CLPP^T145P/WT, and donor-control CLPP^WT/WT fibroblasts (mean ± SD, n = 5 biological replicates). e Patient-derived fibroblasts were treated for 96 h with 2′,3′-dideoxycytidine (ddC) to deplete mitochondrial DNA and steady-state protein levels were analyzed by immunoblotting (mean ± SD, n = 5 biological replicates). Statistical significance was determined for (d, e) by one-way ANOVA with Bonferroni’s post-hoc test (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001). From left to right, p-values are: d  <0.0001, <0.0001, <0.0001, <0.0001, 0.024, 0.0498, <0.0001, 0.0005, 0.0004, <0.0001, 0.003, e 0.023, 0.026. Source data are provided in the Source Data file.

Additionally, it was found that the protein abundance of both the ClpP-interacting AAA+ ATPase ClpX and PDIP38, a protein that interacts with the ZBD of ClpX and regulates its stability¹⁰, were increased in patient CLPP^T145P/T145P cells relative to CLPP^WT/WT or CLPP^T145P/WT cells (Fig. 2a, Supplementary Data 1). Interestingly, PDIP38 has also been suggested to act as an adaptor that modulates ClpX substrate specificity, although no substrates have been proposed to date¹⁰. Thus, to identify putative ClpXP-PDIP38 substrates, we compared our mass spectrometry results with those on BioGRID³⁰, an online public database of genetic and protein interactions, and identified the mitochondrial proteins CHCHD2, ALAS1, and TFAM as likely PDIP38-interacting candidates. Functionally, CHCHD2 is a bi-organellar regulator of oxidative phosphorylation and mitochondrial stress responses³¹ that previously has been found to be upregulated in both Clpp^−/− mouse embryonic fibroblasts and in Clpx^−/− murine hepatocytes, as well as in PRLTS3 patient-derived skin fibroblasts expressing an E221K-ClpP variant^32,33. ALAS1 catalyzes the rate-limiting step in heme biosynthesis³⁴, and while ALAS1 has been previously suggested as a ClpXP substrate³⁵, ClpXP-mediated degradation of ALAS1, and regulation thereof by PDIP38, have yet to be confirmed. Lastly, PDIP38 has been implicated as a potential regulator of TFAM, a mitochondrial transcription factor and critical regulator of mitochondrial DNA (mtDNA) replication and transcription^36–38. Importantly, the abundance of both CHCHD2 and ALAS1 was significantly elevated in CLPP^T145P/T145P patient cells relative to either WT or CLPP^T145P/WT cells, though TFAM levels were unaltered (Fig. 2a, Supplementary Data 1).

Subsequently, gene set enrichment analysis (GSEA) was performed on total hits from the mass spectrometry data for either CLPP^T145P/T145P vs CLPP^WT/WT (Fig. 2b, Supplementary Fig. 2b) or CLPP^T145P/T145P vs CLPP^T145P/WT (Fig. 2c, Supplementary Fig. 2c) whereby the top ten gene ontology (GO) terms and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways were selected based on a q-value cut-off of q ≤ 0.15³⁹, where q-value is the false discovery rate of a statistically significant result. Overall, when comparing mitochondrial proteins (Fig. 2b,c), GO term enrichment was found for biological processes involving peptide and amide biosynthesis and protein metabolism, alongside cellular compartments important to these processes, including ribonucleoprotein complexes and the mitochondrial matrix. Interestingly, KEGG analysis showed that genes related to multiple neurodegenerative diseases were decreased in CLPP^T145P/T145P cells relative to both CLPP^WT/WT and CLPP^T145P/WT cells, suggesting that ClpP may positively regulate processes that normally confer protection against neurodegenerative diseases. Lastly, we found molecular GO terms related to both NADPH oxidoreductase activity and active transmembrane transporter activity were decreased in CLPP^T145P/T145P cells, suggesting impairment in mitochondrial oxidative phosphorylation, a result consistent with prior observation that ClpXP regulates mitochondrial bioenergetics^40,41.

Next, to validate the above findings, immunoblotting was performed on patient-derived fibroblasts, confirming that the steady-state protein levels of ClpX, PDIP38, CHCHD2, and ALAS1 were all significantly increased in CLPP^T145P/T145P fibroblasts relative to CLPP^WT/WT cells, while TFAM remained unchanged (Fig. 2d). Interestingly, the endogenous levels of ClpP were significantly decreased in both the CLPP^T145P/WT and CLPP^T145P/T145P fibroblasts, with ClpP being undetected by immunoblotting from the CLPP^T145P/T145P patient cells, a result that is reflected in our mass spectrometry data by the absence of ClpP peptides from CLPP^T145P/T145P cells (Fig. 2a,d, Supplementary Data 1). Of significance, these endogenous changes in T145P-ClpP protein levels were not due to changes in CLPP mRNA levels (Supplementary Fig. 2d), supporting our earlier results using heterologously-expressed FLAG-tagged T145P-ClpP in HEK293 T-REx cells (Fig. 1b).

While steady-state levels of TFAM were unaltered in CLPP^T145P/T145P patient fibroblasts, it is important to note that the degradation of TFAM may be temporally controlled, such as during spermatogenesis when mtDNA is eliminated⁴². TFAM is a mtDNA-binding protein that modulates mtDNA packaging into mitochondrial nucleoids in a process analogous to histone-chromatin binding in the nucleus⁴³ and the interaction with mtDNA has been shown to protect TFAM from Lon-mediated degradation⁴⁴. To shift the equilibrium towards free, mtDNA-unbound TFAM, and subsequently test whether mtDNA-bound TFAM may likewise be inaccessible to ClpXP, mtDNA was depleted in CLPP^T145P/T145P patient fibroblasts using 2′,3′-dideoxycytidine (ddC), a DNA chain-terminating nucleoside analog that inhibits mtDNA replication⁴⁵. Depletion of mtDNA by ddC was confirmed by qPCR demonstrating a greater than 40% reduction in the mitochondrial genes ATP6, ND1, and CO1 in CLPP^WT/WT, CLPP^T145P/WT, and CLPP^T145P/T145P fibroblasts (Supplementary Fig. 2e). Importantly, ddC treatment induced a decrease in TFAM protein levels, with this effect being greater in both CLPP^WT/WT and CLPP^T145P/WT cells relative to CLPP^T145P/T145P fibroblasts (Fig. 2e). As a control, RT-qPCR confirmed that the change in TFAM steady-state protein levels was not due to a change in TFAM mRNA levels upon ddC treatment (Supplementary Fig. 2f), thus providing evidence for TFAM as a putative ClpXP substrate upon mtDNA depletion.

Validating ClpXP substrates in HEK293 T-REx CLPP^−/− cells

Based on our mass spectrometry analysis, BioGRID search, literature-based evidence, and immunoblotting results we sought to further characterize CHCHD2, ALAS1, and TFAM as putative ClpXP substrates, and to assess whether PDIP38 acts as an adaptor for ClpXP-dependent degradation of these proteins. To begin, a transient siRNA knockdown of ClpP or PDIP38 was performed in HEK293 T-REx cells to investigate whether loss of these proteins influences substrate levels. In support of our analysis from CLPP^T145P/T145P patient fibroblasts (Fig. 2d), siRNA knockdown of ClpP led to an elevation in the steady-state protein levels of ClpX, PDIP38, CHCHD2, and ALAS1 (Fig. 3a). Importantly, siRNA knockdown of PDIP38 led to an increase in CHCHD2 and ALAS1, suggesting a role for PDIP38 as an adaptor in the ClpXP-dependent degradation of these proteins. Knockdown of PDIP38 also led to a loss in ClpX protein, a result mirroring those made by Strack et al. who proposed that PDIP38 protects ClpX from Lon-mediated proteolysis¹⁰.

Fig. 3. Validation of ClpXP substrates in HEK293 T-REx cells.

Steady-state protein levels in HEK293 T-REx cells following a siRNA knockdown of ClpP or PDIP38 and in b WT or CLPP^−/− knockout cells (mean ± SD, n = 5 biological replicates). c Assessment of ClpXP substrate protein solubility by Triton X-100 fractionation in HEK293 T-REx cells following lysis in 1% Triton X-100 buffer. Cells were fractionated into Triton-soluble and insoluble proteins. Triton-insoluble proteins were solubilized in 2×Laemmli sample buffer prior to SDS-PAGE (mean ± SD, n = 4 biological replicates). d HEK293 T-REx cells were treated for 96 h with ddC to deplete mitochondrial DNA and steady-state protein levels were analyzed by immunoblotting (mean ± SD, n = 3 biological replicates). e Cycloheximide chase assay in HEK293 T-REx cells. Protein half-life (T_1/2) of CHCHD2 and ALAS1 were determined by densitometry (ND = non-determined, mean ± SD, n = 3 biological replicates; Supplementary Fig. 3d). f Cycloheximide chase assay in HEK293 T-REx cells following mitochondrial DNA depletion with 100 µM ddC for 96 h. Protein half-life (T_1/2) of TFAM was determined by densitometry (ND = non-determined, mean ± SD, n = 4 biological replicates; Supplementary Fig. 3e). g Submitochondrial fractionation from HEK293 T-REx cells. Isolated crude mitochondria were incubated in either isotonic (intact mitochondria), hypotonic (mitoplast), or Triton X-100 buffer (solubilized mitochondria) in the presence or absence of proteinase K (PK). TOM20 (outer mitochondrial membrane), TIM23 (inner mitochondrial membrane), and Lon (matrix) were used as localization markers (n = 4 biological replicates). h Endogenous co-immunoprecipitation (co-IP) of ClpP, ClpX, CHCHD2, ALAS1, and TFAM with PDIP38 from HEK293 T-REx cells. α-PDIP38 antibody was omitted from control samples (no IgG) and whole-cell lysates (WCL) were prepared alongside co-IP samples (n = 2 biological replicates). Statistical significance was determined for a by one-way ANOVA with Bonferroni’s post-hoc test, (b, c) student’s two-tailed t-test, or (d) by two-way ANOVA with Bonferroni’s post-hoc test (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001). From left to right, p-values are: a 0.019, 0.005, 0.0002, <0.0001, 0.0024, 0.042, 0.004, <0.0001, b <0.0001, <0.0001, 0.0007, <0.0001, c 0.0007, 0.0003, d 0.002, 0.0038, 0.042, 0.0007, 0.021. Source data are provided in the Source Data file.

Next, HEK293 T-REx CLPP^−/− knockout cells were used to further confirm whether CHCHD2, ALAS1, and TFAM are ClpXP substrates. Following immunoblotting, ClpX, PDIP38, CHCHD2, and ALAS1 protein levels were all found to be significantly elevated in the CLPP^−/− knockout cells when compared to WT cells (Fig. 3b). Importantly, the steady-state elevation of these proteins is not a transcriptional response to ClpP depletion as determined by RT-qPCR (Supplementary Fig. 3a), suggesting that these substrate proteins are stabilized in the absence of ClpP. In addition to housekeeping functions, ClpXP has been previously implicated in the degradation of misfolded proteins^40,41, and thus the solubility of these putative ClpXP substrates was measured in HEK293 T-REx cells by performing a Triton X-100 fractionation and separating the Triton-soluble from Triton-insoluble fraction. When comparing WT to CLPP^−/− cells, a significant increase in ALAS1 protein was observed in the Triton-soluble fraction, whereas CHCHD2 accumulated in the Triton-insoluble fraction (Fig. 3c), suggesting that ClpXP may act as a quality control mechanism for misfolded CHCHD2.

Similar to that observed in CLPP^T145P/T145P patient fibroblasts, while TFAM steady-state levels were unaltered following loss of ClpP in HEK293 T-REx cells (Fig. 3a,b), mtDNA depletion by ddC treatment induced a decrease in TFAM protein levels, with this effect being greater in WT cells relative to CLPP^−/− cells (Fig. 3d). Depletion of mtDNA by ddC was confirmed by qPCR demonstrating a greater-than 80% reduction in the mitochondrial genes ATP6, ND1, and CO1 in both WT and CLPP^−/− HEK293 T-REx cells (Supplementary Fig. 3b). As a control, RT-qPCR confirmed that TFAM mRNA levels were not affected by ddC treatment (Supplementary Fig. 3c).

Our steady-state results suggest that ClpXP may regulate the degradation of CHCHD2, ALAS1, and TFAM. Thus, a cycloheximide chase assay was performed to determine the cellular protein half-life (T_1/2) of these proteins over an 8 h chase period in HEK293 T-REx cells. By immunoblotting, enhanced protein stability was observed in CLPP^−/− cells for both CHCHD2 and ALAS1 (Fig. 3e, Supplementary Fig. 3d). The stability of TFAM was unchanged in vehicle-treated cells; however, following depletion of mtDNA by ddC treatment, stabilization of TFAM was observed in CLPP^−/− cells when compared to WT cells (Fig. 3f, Supplementary Fig. 3e). No turnover of either ClpX or PDIP38 was observed over the 8 h cycloheximide chase period in either WT or CLPP^−/− cells suggesting that both proteins are highly stable and long-lived (Fig. 3e,f, Supplementary Fig. 3d,e).

Next, we tested whether these proteins may interact together in cells. It has previously been shown that ClpXP¹⁰, PDIP38¹⁰, ALAS1^46,47, and TFAM^48,49 localize to the mitochondrial matrix, however, CHCHD2 is known to predominantly localize to the intermembrane space (IMS) of the mitochondria⁵⁰. Therefore, to confirm that these proteins co-localize to the same mitochondrial sub-compartment, we performed a submitochondrial fractionation (Fig. 3g) from isolated mitochondria (Supplementary Fig. 3f). Importantly, by immunoblotting we confirmed that ClpXP, PDIP38, and TFAM all localize to the mitochondrial matrix in HEK293 T-REx cells with ALAS1 exhibiting mixed IMS/matrix localization. Interestingly, while CHCHD2 appears to localize exclusively to the IMS in WT cells, it accumulates in the mitochondrial matrix in CLPP^−/− cells, suggesting that under physiological conditions a small proportion of CHCHD2 may transit into the mitochondrial matrix where it is degraded by ClpXP. Subsequently, an anti-PDIP38 co-immunoprecipitation was performed from HEK293 T-REx cells, demonstrating complex formation of PDIP38 with ClpP, ClpX, CHCHD2, ALAS1, and TFAM (Fig. 3h).

Validating ClpXP substrates in Perrault-related C144R-ClpP (c.403 T > C) stem cells

We next sought to confirm CHCHD2, ALAS1, and TFAM as putative ClpXP substrates in a Perrault syndrome model using induced pluripotent stem cells (iPSCs) and iPSC-derived neural progenitor cells (NPCs). To accomplish this, two isogenic CRISPR-engineered iPSC lines (PGP1 cells; clones H1 and H3) were generated harboring a homozygous Perrault-related CLPP^C144R/C144R (c.403 T > C) variant. In contrast to T145P-ClpP, the C144R variant did not alter the steady-state levels of ClpP in HEK293 T-REx cells (Fig. 1b), suggesting that an alternative mechanism beyond ClpP protein expression is responsible for dysfunction of this ClpP variant in PRLTS3. Once generated, successful genomic mutagenesis of ClpP on residue Cys¹⁴⁴ was confirmed by Sanger sequencing, and karyotype analysis demonstrated that no chromosomal abnormalities were introduced into the C144R-ClpP iPSC line following CRISPR editing (Supplementary Fig. 4a). Furthermore, by RT-qPCR (Supplementary Fig. 4b) and immunofluorescence microscopy (Fig. 4a), C144R-ClpP iPSCs were found to be positive for the pluripotency markers Oct4, Sox2, Nanog, and SSEA-4, and, importantly, that there were no observable differences in these markers between WT and C144R-ClpP iPSCs. Subsequently, iPSC-derived NPCs were generated from these cells using a commercial kit based on the dual-SMAD inhibition approach for neural induction (Fig. 4b)^51,52. We chose to differentiate iPSCs into NPCs on the basis that Perrault syndrome pathology is characterized by sensorineural hearing loss and is often accompanied by leukodystrophy and related neuropathology⁵³, making NPCs a disease-relevant model appropriate for follow-up analysis. Successful neural induction was validated by immunofluorescence microscopy, where both WT and C144R-ClpP NPCs were shown to be positive for the expression of NPC markers Pax6, Nestin, and Sox1 (Fig. 4c).

Fig. 4. Validation of ClpXP substrates in Perrault-related C144R-ClpP (c.403 T > C) stem cells.

a Immunofluorescent staining of WT and C144R-ClpP (clones H1 and H3) iPSCs for the pluripotency markers Oct4 (nuclear), SSEA-4 (cell surface), Sox2 (nuclear), and Nanog (nuclear). Nuclei were counterstained with DAPI. Scale bar = 250 μm, n = 3 (Sox2 and Nanog) or n = 5 (SSEA-4 and Oct4) biological replicates. b Overview of neural induction protocol for iPSC-derived neural progenitor cells (NPCs). c Immunofluorescent staining of iPSC-derived NPCs for the neuroectodermal markers Pax6 (nuclear), Nestin (cytosolic), and Sox1 (nuclear). Nuclei were counterstained with Hoechst. Scale bar = 250 μm, n = 5 (Sox1) and n = 6 (Pax6 and Nestin) biological replicates. d, e Steady-state protein levels in iPSCs (d) and NPCs (e) (mean ± SD, n = 5 biological replicates). iPSCs (f) and NPCs (g) were treated for 96 h with ddC to deplete mitochondrial DNA and steady-state protein levels were analyzed by immunoblotting (mean ± SD, iPSCs n = 4, NPCs n = 3 biological replicates). Cycloheximide chase assay in iPSCs (h) and NPCs (i). Protein half-life (T_1/2) of ALAS1 was determined by densitometry (ND = non-determined, n = 3 biological replicates; Supplementary Fig. 5g,h). Statistical significance was determined for (d–g) by one-way ANOVA with Bonferroni’s post-hoc test (*p ≤ 0.05, **p ≤ 0.01, ***p ≤ 0.001, ****p ≤ 0.0001). From left to right, exact p-values are: d 0.0045, 0.0027, 0.0008, 0.016, <0.0001, <0.0001, e <0.0001, <0.0001, <0.0001, <0.0001, 0.0069, 0.016, <0.0001, <0.0001, f 0.0068, 0.026, g 0.0014, 0.0029. Source data are provided in the Source Data file.

Immunoblotting of ClpP from whole cell lysates showed that the endogenous steady-state protein levels of WT and C144R-ClpP were similar in both iPSCs and NPCs, and that there were no significant differences when comparing the two C144R-ClpP clones H1 and H3 (Fig. 4d,e). However, similar to both CLPP^T145P/T145P patient fibroblasts (Fig. 2d) and HEK293 T-REx CLPP^−/− cells (Fig. 3b), the protein levels of ClpX, PDIP38, and ALAS1 were increased in both C144R-ClpP iPSCs (Fig. 4d) and C144R-ClpP NPCs (Fig. 4e) when compared to WT cells. CHCHD2 was also elevated in the C144R-ClpP NPCs (Fig. 4e); however, CHCHD2 protein was virtually absent in both WT and C144R-ClpP iPSCs (Supplementary Fig. 4c), a result consistent with previous observations made in embryonic stem cells and human iPSCs⁵⁴. Importantly, the mRNA levels for these proteins were unaffected in both C144R-ClpP iPSCs and NPCs (Supplementary Fig. 5a,b), suggesting that steady-state elevation of these proteins is not a transcriptional response to ClpP dysfunction.

We next tested whether TFAM is a ClpXP substrate in our stem cell models as observed in CLPP^T145P/T145P patient fibroblasts (Fig. 2e) and HEK293 T-REx CLPP^−/− cells (Fig. 3d). Importantly, while protein levels of TFAM remained unchanged under steady-state conditions in both the C144R-ClpP iPSCs (Fig. 4d) and C144R-ClpP NPCs (Fig. 4e), depletion of mtDNA in iPSCs and NPCs by ddC treatment led to a loss in TFAM protein levels that was greater in WT cells relative to C144R-ClpP cells without affecting TFAM mRNA levels (Fig. 4f,g, Supplementary Fig. 5c-e). Successful depletion of mtDNA was confirmed by qPCR showing a >50% reduction in the mitochondrial genes ATP6, ND1, and CO1 in both WT and C144R-ClpP iPSCs and NPCs (Supplementary Fig. 5d,f).

Subsequently, cycloheximide chase assays were carried out to determine the turnover of the putative ClpXP substrates in the Perrault-related iPSCs (for 8 h) and NPCs (for 24 h). Of these, ALAS1 was found to be stabilized in both C144R-ClpP iPSCs (Fig. 4h, Supplementary Fig. 5g) and C144R-ClpP NPCs (Fig. 4i, Supplementary Fig. 5h). No turnover of ClpX, PDIP38, TFAM or CHCHD2 was observed over the cycloheximide chase periods in WT or C144R-ClpP iPSCs/NPCs suggesting that these proteins are highly stable and long-lived in these cells (Fig. 4h,i, Supplementary Fig. 5g,h). Lastly, while TFAM appears to be stabilized following ddC treatment in C144R-ClpP iPSCs and NPCs compared to WT cells (Fig. 4f,g), we were unable to capture turnover of TFAM over the cycloheximide chase periods in either WT or C144R-ClpP iPSCs/NPCs following ddC-induced mtDNA depletion (Supplementary Fig. 5i,j). Given these results, it is likely that the half-life of TFAM exceeds the cycloheximide treatment window in ddC-treated iPSCs and NPCs.

PDIP38 enhances the in vitro degradation of CHCHD2, ALAS1, and TFAM by ClpXP

Our results from CLPP^T145P/T145P patient fibroblasts, HEK293 T-REx CLPP^−/− cells, and Perrault-related C144R-ClpP iPSCs and NPCs suggest that CHCHD2, ALAS1, and TFAM are cellular  ClpXP substrates with dysregulated turnover in PRLTS3. Furthermore, we showed that PDIP38, a proposed ClpX adaptor¹⁰, can form a complex with these substrates by co-immunoprecipitation (Fig. 3h), and therefore we wanted to confirm whether these proteins can interact directly in vitro. Hence, recombinant proteins for each substrate were tested for their ability to bind PDIP38 in an in vitro pulldown assay. For all in vitro assays, proteins were expressed and purified from E. coli without their predicted MTS except for CHCHD2, which was used with its full-length sequence. Using purified His₁₂-SUMO-PDIP38 as a bait protein, CHCHD2 (Fig. 5a), ALAS1 (Fig. 5b), and TFAM (Fig. 5c) were found to directly interact with PDIP38 in vitro. As a control, and in agreement with prior evidence¹⁰, we show that PDIP38 and ClpX interact in vitro using a purified His₇-ClpX bait (Fig. 5d).

Fig. 5. Effect of PDIP38 on the ClpXP-dependent degradation of CHCHD2, ALAS1, and TFAM.

In vitro pulldown of purified CHCHD2 (a), ALAS1 (b), or TFAM (c) with His₁₂-SUMO-PDIP38, or pulldown of purified PDIP38 with His₇-ClpX (d). n = 2 (CHCHD2, ALAS1, PDIP38) and n = 4 (TFAM) biological replicates. e In vitro degradation of CHCHD2, ALAS1, and TFAM in the presence or absence of ClpP, ClpX, and PDIP38. Loss of substrates was quantified from Coomassie Brilliant Blue (CBB)-stained SDS-PAGE gels or from anti-ALAS1 immunoblots (IB). Degradation rate constants (k) are shown for conditions where substrates were degraded by ClpXP (mean ± SD, n = 3 technical replicates). f In vitro degradation of TFAM by ClpXP-PDIP38 in the presence or absence of a double-stranded oligonucleotide corresponding to the TFAM binding site upstream of the mtDNA light strand promoter (uLSP) or a scrambled sequence. Degradation of α-casein and CHCHD2 was used as a control to ensure that the presence of DNA did not inhibit ClpXP activity (n = 3 biological replicates). Source data are provided in the Source Data file.

Next, to validate our cellular observations, in vitro degradation assays for the three putative ClpXP substrates were performed using recombinant ClpXP in the presence or absence of purified PDIP38. Generally, CHCHD2, ALAS1, and TFAM were all degraded by ClpXP in vitro but weakly; the addition of PDIP38 significantly enhanced their degradation as denoted by the increase in their respective degradation rate constants (Fig. 5e). As noted earlier, interaction of TFAM with mtDNA, specifically a double-stranded oligonucleotide corresponding to the TFAM binding site upstream of the mtDNA light strand promoter (uLSP)^44,55, has been shown to protect TFAM from Lon-mediated degradation⁴⁴. Interestingly, using the same TFAM-binding mtDNA sequence, as well as a scrambled oligo, we found that the in vitro degradation of TFAM by ClpXP-PDIP38 was likewise abolished in the presence of DNA (Fig. 5f), supporting our observations made in mtDNA-depleted HEK293 T-REx CLPP^−/− cells and C144R-ClpP iPSCs and NPCs (Figs. 3d, 4f,g). Collectively, these in vitro and cell data demonstrate that CHCHD2, ALAS1, and TFAM are bona fide ClpXP substrates and that their degradation by ClpXP is enhanced by PDIP38.

Characterization of Perrault-related ClpP variants reveals different mechanisms of ClpP dysfunction

Comparison of the Perrault-related C144R- and T145P-ClpP variants revealed similar effects on ClpXP substrates in cells, though our steady-state analysis showed differing effects on mutant ClpP protein levels, suggesting that ClpP dysfunction by these mutations occurs through alternative mechanisms. Therefore, using in vitro recombinant purified proteins, we next sought to biochemically characterize C144R- and T145P-ClpP, alongside several other PRLTS3-related ClpP variants that did not affect the steady-state expression of ClpP when transiently expressed in HEK293 T-REx cells (Fig. 1b). By size exclusion chromatography (SEC), WT-, T135S-, C144R-, C147S-, I208M-, A210T-, and Y229D-ClpP appear to form a tetradecameric oligomer, with Y229D-ClpP also displaying a notable peak in lower molecular weight fractions (Supplementary Fig. 6a). T145P-ClpP exclusively eluted as a higher molecular weight soluble aggregate by SEC. To further assess these findings in cells, FLAG-tagged WT-, C144R-, and T145P-ClpP were transiently expressed in HEK293 T-REx CLPP^−/− cells and a Triton X-100 fractionation was performed. Following immunoblotting, T145P-ClpP was found to accumulate in the Triton-insoluble fraction, whereas WT- and C144R-ClpP were both found predominantly in the Triton-soluble fraction (Supplementary Fig. 6b). Furthermore, T145P-ClpP was found to form SDS-resistant higher molecular weight species in the Triton-insoluble fraction, suggesting that this variant aggregates in cells.

Subsequently, the in vitro peptidase activities of these purified Perrault-related ClpP variants were tested using an AMC-conjugated N-acetyl-Trp-Leu-Ala (WLA) peptide as a substrate⁵⁶ and in the presence or absence of ClpX, as ClpX is known to enhance the peptidase activity of human ClpP⁵⁷ (Fig. 6a). This revealed four different outcomes (Fig. 6a, Supplementary Fig. 6c): (1) ClpP is active and degradation of WLA-AMC was enhanced by ClpX (WT, C147S), (2) ClpP is active but addition of ClpX had no further stimulatory effect (T135S, C144R), (3) ClpP is inactive but peptidase activity was restored by ClpX (A210T), and (4) ClpP is inactive both alone and with ClpX (T145P, I208M, and Y229D). While ClpP does not require ClpX to degrade peptides, the ATP-dependent unfoldase activity of ClpX is required for ClpP-mediated proteolysis of folded proteins¹³. Therefore, we next performed in vitro ClpXP substrate degradation assays using folded protein substrates in the presence of ClpX alone (α-casein) or with ClpX and PDIP38 (CHCHD2, ALAS1, or TFAM). All tested ClpP variants displayed a decrease in the degradation rates for these substrates relative to WT-ClpP (Fig. 6b,c, Supplementary Fig. 6d), with C147S-ClpP and A210T-ClpP showing the least decrease. These results suggest that the turnover of folded proteins by these Perrault-related ClpP variants in the presence of ClpX and PDIP38 was overall impaired.

Fig. 6. Analysis of the catalytic activity of Perrault-related ClpP variants.

a In vitro ClpP peptidase assay using the substrate WLA-AMC to determine the catalytic activity of WT and Perrault-variant ClpP proteins in the presence (green) or absence (black) of ClpX (n = 3 technical replicates). Initial rates were quantified and reported in Supplementary Fig. 6c. b, c In vitro degradation of α-casein by ClpXP (b) or of CHCHD2, ALAS1, or TFAM by ClpXP with PDIP38 (c). Loss of substrates was visualized on Coomassie Brilliant Blue (CBB)-stained SDS-PAGE gels or by anti-ALAS1 immunoblotting (IB) (mean ± SD, n = 3 biological replicates). Relative rate constants are reported in Supplementary Fig. 6d. Source data are provided in the Source Data file.

Given that analysis of C144R-ClpP revealed defects in the degradation of ClpXP substrates both endogenously in iPSCs and NPCs as well as in vitro, we decided to characterize this mutation in further detail. Residue C144 is located within the ClpX-binding hydrophobic pocket of ClpP (Fig. 1a), and since the addition of ClpX had no effect on the in vitro peptidase activity of C144R-ClpP (Fig. 6a), we postulated that the C144R mutation may interfere with the interaction between C144R-ClpP with ClpX. Indeed, co-immunoprecipitation of endogenous ClpX with either transiently expressed FLAG-tagged C144R-ClpP in HEK293 T-REx CLPP^−/− cells (Fig. 7a) or with endogenous C144R-ClpP in Perrault-related NPCs (Fig. 7b) was reduced when compared to WT-ClpP. Furthermore, by in vitro pulldown with purified His₇-ClpX, the interaction of ClpX with recombinant C144R-ClpP was significantly diminished relative to WT-ClpP (Fig. 7c).

Fig. 7. Molecular characterization of the C144R-ClpP Perrault-related variant.

a Co-immunoprecipitation (co-IP) of endogenous ClpX with transiently expressed 3×FLAG-tagged ClpP in HEK293 T-REx CLPP^−/− cells. Control cells were transfected with empty vector and whole-cell lysates (WCL) were prepared alongside co-IP samples (mean ± SD, n = 3 biological replicates). b Co-IP of endogenous ClpX with endogenous ClpP in WT or Perrault-related C144R-ClpP NPCs. α-ClpP antibody was omitted in control samples (no IgG; mean ± SD, n = 3 biological replicates). c In vitro pulldown of purified ClpP with His₇-ClpX (mean ± SD, n = 3 biological replicates). d Surface representation of the C144R-ClpP tetradecameric crystal structure (PDB: 9PN7) solved in a compact conformation showing a partially obstructed axial pore. Mutant C144R residues are shown (red spots) on the surface of ClpP. Surface representations centered on the ClpX LGF loop binding site of a ClpP hydrophobic pocket of either C144R-ClpP (e) or human WT-ClpP (f; PDB: 9DKV) in the compact confirmation. Each hydrophobic pocket is formed by two adjacent ClpP protomers, where residues Y138 and W146, in combination with residues L104′ and Y118′ from a neighboring ClpP protomer, form the LGF loop’s binding site⁸. g Superposition of the hydrophobic pockets of C144R-ClpP (e; green cartoon) and WT-ClpP (f; gray cartoon) in the compact confirmation. Residue side chains are shown as sticks. The distance between the Cα atom of WT-Cys¹⁴⁴ or mutant Arg¹⁴⁴ and the Gln^107′ Cα atom of the adjacent protomer within a hydrophobic pocket was measured (dashed line) and was found to be 18.7 Å and 18.3 Å, respectively, demonstrating an inwards compaction as indicated by the arrows. Statistical significance was determined for (a, c) by student’s two-tailed t-test or b by one-way ANOVA with Bonferroni’s post-hoc test (***p ≤ 0.001, ****p ≤ 0.0001). From left to right, p-values are: a <0.0001, b 0.0002, 0.0001, c <0.0001. Source data are provided in the Source Data file.

Subsequently, the X-ray crystal structure of C144R-ClpP was solved (PDB: 9PN7; Table 1). The C144R-ClpP tetradecamer was found in the compact conformation similar to that observed for other complexes of human ClpP (Fig. 7d, Supplementary Fig. 7a)^56,58. Importantly, we saw a narrowing of the surface-exposed ClpX-binding hydrophobic pocket of C144R-ClpP (Fig. 7e) when compared to that of WT-ClpP in the compact conformation (Fig. 7f,g). Furthermore, the Arg¹⁴⁴ mutation increased electropositivity of the hydrophobic pocket through the introduction of a charged residue (Supplementary Fig. 7b). Thus, the combination of decreased ClpX-binding pocket size, altered surface potential, and the several rotameric states of the bulky arginine side chain observed between different ClpP protomers within the C144R-ClpP structure (Supplementary Fig. 7c) are predicted to interfere with the binding of ClpX to C144R-ClpP (Supplementary Fig. 7d), weakening ClpXP complex formation and, as a result, reducing ClpXP-dependent degradation of folded substrates.

Table 1.

X-ray data collection and refinement statistics

	C144R-ClpP apo	C144R-ClpP + TR-65
PDB ID	9PN7	9PNC
Resolution Range (Å)	44.11-3.02 (3.12-3.02)	49.44–2.68 (2.75–2.68)
Space group	C 1 2 1	C 1 2 1
a, b, c (Å) α, β, γ (°)	142.16, 146.56, 102.05 90, 90, 90	181.23, 148.50, 151.33 90, 101.44, 90
Unique reflections	34,157 (2,588)	107,617 (6,419)
Multiplicity	6.7 (6.6)	5.9 (5.4)
Completeness (%)	96.10 (73.37)	99.6 (99.0)
Mean I/sig(I)	5.4 (0.56)	7.4 (0.52)
R-merge	0.250 (2.047)	0.177 (2.228)
R-meas	0.271 (2.221)	0.194 (2.466)
R-pim	0.103 (0.855)	0.079 (1.036)
CC1/2	0.988 (0.391)	0.997 (0.390)
Reflections used in refinement	34,086 (2,582)	107,617 (6,419)
Reflections used for R-free	2000 (152)	1995 (119)
R-work	0.2233 (0.3447)	0.2369 (0.3859)
R-free	0.2563 (0.3532)	0.2758 (0.3976)
No. of non-hydrogen atoms	9511	18,941
macromolecules	9,463	18,265
ligands	0	434
solvent	48	242
Protein residues	1207	2339
RMS (bonds) (Å)	0.002	0.006
RMS (angles) (°)	0.44	1.08
Ramachandran favored (%)	95.97	96.71
Ramachandran allowed (%)	3.95	2.93
Ramachandran outliers (%)	0.09	0.35
Rotamer outliers (%)	0.00	1.18
Clashscore	6.75	15.97
Average B-factor (Å2)	77.45	62.90
macromolecules	77.53	63.05
ligands	Not applicable	61.70
solvent	60.79	53.29

Statistics for the highest resolution shells are found in parentheses.

Degradation of protein substrates by C144R-ClpP can be partially restored with the ClpP-activating drug TR-65

Based on the above analysis of the C144R-ClpP variant, our data suggest that one mechanism for ClpP dysfunction in Perrault syndrome is due to impaired ClpXP complex formation and the resultant accumulation of ClpXP substrates. Therefore, we next investigated whether impaired substrate degradation can be partially restored through pharmacological activation of Perrault-variant ClpP, such as with acyldepsipeptides (ADEPs) or imipridone-based compounds (e.g. ONC201 or TR-65)^58–60. Briefly, these compounds have been shown to bind within the ClpP hydrophobic pocket displacing ClpX, opening the ClpP axial pores, and as a result allowing for folded proteins to freely enter into the ClpP catalytic core in a ClpX-independent manner, resulting in nonspecific protein degradation^12,56. Using α-casein as a model substrate, an in vitro ClpP degradation assay was performed in the presence of TR-65 demonstrating that, in the absence of ClpX, C144R-ClpP but not T145P-ClpP can be activated to degrade α-casein similar to that observed with WT-ClpP (Fig. 8a). Michaelis-Menten analysis using FITC-labeled casein revealed that both the k_cat and K_M for the protease activity of C144R-ClpP were greater than that of WT-ClpP when activated with TR-65 (Fig. 8b). Thus, while the peptidase activity of apo-C144R-ClpP appeared unaffected (Fig. 6a, Supplementary Fig. 6c), positioning of the C144R mutation within the ClpX-binding hydrophobic pocket may have a positive allosteric effect on ClpP activation by TR-65, a mechanism that has been documented with ONC201 and human ClpP⁶⁰, for example, among other compounds.

Fig. 8. Effect of TR-65 on C144R-ClpP proteolytic activity.

a In vitro ClpP degradation of α-casein in the presence or absence of TR-65. Loss of protein was quantified from Coomassie-stained SDS-PAGE gels. Degradation rate constants (k) are given where α-casein were degraded by ClpP (mean ± SD, n = 3 technical replicates). b In vitro ClpP proteolytic assay using the substrate FITC-casein was used to determine the catalytic activity of WT- and C144R-ClpP in the presence or absence of TR-65. The curve was fitted to the Michaelis-Menten equation and turnover rate (k_cat), Michaelis constant (K_M), and catalytic efficiency (k_cat / K_M) were calculated (mean ± SD, n = 3 technical replicates). K_d,_app was calculated from Supplementary Fig. 8a,b. c Surface representation of the solved co-crystal structure of tetradecameric C144R-ClpP bound to TR-65 (PDB: 9PNC). The structure was solved in a compact conformation showing an enlarged axial pore. Mutant C144R residues are shown on the surface of ClpP (red spots). Bound TR-65 molecules are shown as light blue sticks. d TR-65 binds to the C144R-ClpP hydrophobic pocket formed between residues Y138 and W146, in combination with residues L104′ and Y118′ from a neighboring ClpP protomer⁸ (left panel). The m-nitrile substituted benzyl group of TR-65 mimics the binding of the Phe residue of the ClpX LGF loop within the ClpP hydrophobic pocket. An alternate view is presented (right panel) showing only select residue side chains around TR-65 and the mutant Arg¹⁴⁴ residue located apical of the ClpP-bound TR-65 within the ClpP hydrophobic pocket. The C144R-ClpP surface and stick representation are colored green, with surface and side chains of some amino acid residues and of TR-65 (light blue sticks) colored according to atom to show polarity: N; blue, O; red. e In vitro ClpP degradation of TFAM, and CHCHD2 in the presence or absence of TR-65. Loss of protein was quantified from Coomassie-stained SDS-PAGE gels. Degradation rate constants (k) are given where TFAM or CHCHD2 were degraded by ClpP (mean ± SD, n = 3 technical replicates). f, g Steady-state protein levels in WT or Perrault-related C144R-ClpP iPSCs f and NPCs g treated with TR-65 for 24 h (n = 3 biological replicates). Source data are provided in the Source Data file.

Subsequently, in vitro dose-titration experiments were performed by varying ClpP and TR-65 concentrations to obtain apparent binding constants (K_d,app; Supplementary Fig. 8a,b). While the affinity of TR-65 to C144R-ClpP was slightly reduced compared to WT-ClpP, the apparent dissociation constants for both proteins were in the low nanomolar range (Fig. 8b, Supplementary Fig. 8a,b), suggesting tight binding of TR-65 to C144R-ClpP. In support, the crystal structure of C144R-ClpP in complex with TR-65 was solved (PDB: 9PNC; Table 1). The TR-bound C144R-ClpP tetradecamer was found in the same compact conformation as the apo-C144R-ClpP, albeit with widened axial pores indicative of ClpP activation (Fig. 8c, Supplementary Fig. 8c,d)^12,61. In the ClpP structure, the mutant Arg¹⁴⁴ residue is positioned apical of the hydrophobic pocket and does not occlude TR-65 binding (Supplementary Fig. 8e), though it is likely that the positively charged arginine side chain would reduce the binding affinity of C144R-ClpP for TR-65 (Fig. 8d). Of significance, structural analysis showed that TR-65 bound similarly within the hydrophobic pocket of WT- and C144R-ClpP (Supplementary Fig. 8f,g), and in a manner consistent with the reported effects of TR-65 on human ClpP^12,58. To further assess whether the mutant Arg¹⁴⁴ residue may impact the ability of C144R-ClpP to be activated in vitro, we used a bulkier acyldepsipeptide-based ClpP activator, ADEP-28⁵⁶. Importantly, proteolytic activity (Supplementary Fig. 8h) and structural docking analysis using the solved crystal structure of WT-ClpP bound to ADEP-28 (Supplementary Fig. 8i) indicated that, like TR-65, ADEP-28 was able to activate C144R-ClpP in a manner similar to WT-ClpP (Supplementary Fig. 8h,i). Overall, these data suggest that the binding of these small molecule activators to the hydrophobic pocket of ClpP is not hindered by the presence of the mutant residue Arg¹⁴⁴.

Having established that TR-65 can activate C144R-ClpP to degrade casein in vitro, we next examined whether TR-65 can likewise induce degradation of the identified ClpXP substrates by C144R-ClpP. First, using purified proteins, we were able to successfully observe the ClpX-independent degradation of both CHCHD2 and TFAM in vitro by activated ClpP, though we saw no degradation of ALAS1 by either WT- or C144R-ClpP when treated with TR-65 (Fig. 8e, Supplementary Fig. 8j). Similar to α-casein (Fig. 8a), the rates of TR-65-induced degradation of CHCHD2 and TFAM were slightly elevated for C144R-ClpP relative to WT-ClpP as denoted by the rate constants (Fig. 8e). Subsequently, to determine whether TR-65 can activate endogenous C144R-ClpP in cells, we performed immunoblotting from Perrault-related C144R-ClpP iPSCs and NPCs treated with TR-65. Following a 24 h treatment with increasing doses of TR-65, the steady-state protein levels of ALAS1 and TFAM were reduced in both WT and C144R-ClpP iPSCs (Fig. 8f) and NPCs (Fig. 8g). These results agree with previously published data in other cell lines showing that ClpP activation by the related imipridones TR-57, TR-27, and ONC201, as well as the acyldepsipeptide ADEP-14, results in a depletion of TFAM protein^58,59,62,63. As mentioned earlier, CHCHD2 was virtually absent in our iPSCs (Supplementary Fig. 4c) and, while detectable in NPCs, CHCHD2 protein levels remained unchanged in both WT and C144R-ClpP NPCs treated with TR-65 (Fig. 8g). Interestingly, for C144R-ClpP cells, higher doses of TR-65 were required to observe loss of substrate protein compared to WT cells (Fig. 8f,g). This dose-dependent effect was also seen for the TR-65-induced reduction in the steady-state protein levels of ClpP, ClpX, and PDIP38 in both iPSCs and NPCs (Fig. 8f,g). Overall, these results demonstrate that the proteolytic activity of Perrault-related C144R-ClpP can be restored both in vitro and in cells with the ClpP-activating TR-65 compound, though with limitations with respect to targeted substrates.

Discussion

Many biallelic variants in CLPP are associated with PRLTS3, though few have been extensively characterized in vitro or, importantly, in cells. Our results, alongside prior in vitro data¹⁷, suggest multiple distinct molecular mechanisms by which ClpP mutations cause ClpP dysfunction in PRLTS3. We propose that these mutations can be non-exclusively grouped into those that (1) cause ClpP misfolding and/or loss of mutant ClpP protein (e.g. T145P), (2) impede the ClpP-ClpX interaction (e.g. C144R), (3) interfere in the activation of ClpP by ClpX (e.g. T135S, T145P, C144R, I208M, and Y229D), (4) directly or indirectly affect the ClpP active site (e.g. T145P, C147S, I208M, A210T, and Y229D), and/or (5) alter the oligomerization of ClpP (Y229D). It is important to note that, in addition to missense mutations, frame-shifts, deletions, and compound heterozygous mutations in CLPP have also been observed in PRLTS3 patients^53,64,65. Thus, and of significance, the detailed biochemical and cellular characterization of Perrault-related ClpP variants reported in this study may serve as the basis for future investigation of other ClpP variants in Perrault syndrome.

Identification of ClpP substrates affected in Perrault syndrome was of key interest to our study as bona fide endogenous substrates of human ClpP have been largely unreported. Critically, we present data on three newly identified ClpP substrates in vitro and in cells, namely CHCHD2, ALAS1, and TFAM. Of these, we observed steady-state elevation of both CHCHD2 and ALAS1 in patient-derived CLPP^T145P/T145P fibroblasts, HEK293 T-REx CLPP^−/− cells, and in C144R-ClpP iPSCs and NPCs. Functionally, ALAS1 is a mitochondrial enzyme that catalyzes the initial and rate-limiting step in heme biosynthesis³⁵. Interestingly, ClpX has been shown to independently regulate the activity and heme-dependent degradation of ALAS1^35,66–68. Inactivating ClpX mutations have been associated with the development of erythropoietic protoporphyria, a disease characterized by toxic accumulation of the heme-intermediate protoporphyrin IX in erythroid cells partly due to stabilization of ALAS2, an erythroid-specific ALAS isozyme^69,70. It is currently unclear whether ALAS1 stabilization due to ClpP dysfunction results in elevation of cellular heme or heme intermediates in PRLTS3 patients.

CHCHD2 is a bi-organellar regulator of oxidative phosphorylation and mitochondrial stress responses, playing important roles in regulating electron flow in the mitochondrial ETC and acting as a nuclear transcription factor for both itself and COX4I2, a cytochrome c oxidase subunit, in response to hypoxic stress³¹. While reported to localize to the IMS⁵⁰, we observed that CHCHD2 possesses dual localization within the mitochondria, being found in both the IMS and matrix in HEK293 T-REx CLPP^−/− cells (Fig. 3g). Interestingly, CHCHD2 was only detectable in the IMS in wild-type cells, suggesting that a proportion of this protein may transit into the mitochondrial matrix where it is rapidly degraded by ClpXP under basal conditions. CHCHD2 belongs to a family of coiled-coil-helix-coiled-coil-helix (CHCH) domain-containing proteins that possess CX₉C motifs that are imported into the IMS through the mitochondrial disulfide-relay MIA import pathway^71,72. Recently, NDUFAF8, another CX₉C domain-containing protein, was reported to contain a weak MTS that drives slow TIM23-dependent import of NDUFAF8 into the mitochondrial matrix, allowing for exposure to the IMS disulfide relay system and subsequent oxidative folding of NDUFAF8 while on route to the matrix. Importantly, levels of NDUFAF8 in the matrix were found to be regulated by ClpP, where reduced and potentially misfolded NDUFAF8 was stabilized in CLPP knockout cells⁷³.

CHCHD2 is also proposed to contain a weak MTS⁷³, and based on our observation that CHCHD2 was enriched in the Triton-insoluble fraction of HEK293 T-REx CLPP^−/− cells (Fig. 3c), we propose that ClpXP acts as a quality control mechanism for CHCHD2, and potentially other CX₉C domain-containing proteins, by degrading mislocalized and misfolded CHCHD2 in the mitochondrial matrix. Interestingly, while mitochondrial processing peptidases (MPPs) typically remove the MTS of proteins imported into the mitochondrial matrix^74,75, CHCHD2 appears to retain its MTS in cells (Fig. 3g, Supplementary Fig. 3f)⁷³. Nevertheless, as a control, we also demonstrated that recombinant purified CHCHD2 lacking its putative MTS (residues 1-52) interacted with PDIP38 and is degraded by ClpXP-PDIP38, and by TR-65 activated ClpP in vitro (Supplementary Fig. 9). It is interesting to note that mutations in CHCHD2, many of which are MTS-localized, have been implicated in neurodegenerative disorders including Parkinson’s disease and Lewy body dementia⁷⁶, with one variant, T61I, resulting in mislocalized and aggregated CHCHD2 in patient brain samples^77,78.

Another identified substrate for ClpXP is TFAM. While TFAM is a direct ClpXP substrate in vitro (Fig. 5e), ClpXP-dependent loss of TFAM in cells was only observed when mtDNA was depleted using ddC (Figs. 2e, 3d,f, 4f,g). Additionally, we found that supplementation of recombinant TFAM with double-stranded oligonucleotides matching a known TFAM-mtDNA binding sequence^44,55 prevented the in vitro degradation of the protein by ClpXP (Fig. 5f), suggesting that ClpP-mediated degradation of TFAM in cells may be regulated by binding of TFAM to mtDNA. Turnover of damaged mtDNA under both physiological and cell stress conditions is an essential quality control mechanism that maintains mitochondrial genome integrity^79–82. In addition, elimination of paternal mtDNA during spermatogenesis is critical to ensure the sole inheritance of maternal mtDNA in many sexually reproducing animals⁸³. Intriguingly, two recent studies have reported that loss or disruption of the ClpX adaptor protein PDIP38 leads to increased mtDNA retention in Drosophila spermatids, resulting in dysfunctional spermatids and sterility, and that co-knockdown of ClpX and PDIP38 led to an increase in TFAM protein^42,84. Furthermore, Clpp^−/− mouse models have implicated ClpP in the regulation of mtDNA nucleoid formation and mtDNA stability^32,85; TFAM plays a critical role in both of these functions⁴³. Thus, while future work is needed to investigate the role of human ClpXP in sperm maturation, our data alongside the aforementioned studies suggest that ClpXP regulates paternal mtDNA elimination during spermatogenesis through the degradation of TFAM in a PDIP38-dependent manner. Importantly, and of relevance to the ClpP variants in our work, azoospermia has been reported in a PRLTS3 patient homozygous for the C144R-ClpP (c.403 T > C) variant¹⁸, and thus we propose that dysregulation of the ClpXP-dependent degradation of TFAM plays a critical role in the etiology of PRLTS3.

In addition to the newly identified ClpP substrates, we observed elevation of both ClpX and PDIP38 in both CLPP^−/− cells and in cells endogenously expressing Perrault-related ClpP variants. Interestingly, the increase in ClpX and PDIP38 in these cells was not due to an increase in CLPX or PDIP38 mRNA (Supplementary Fig. 3a, 5a,b), suggesting that these proteins are stabilized when cellular ClpP function is lost. Little is known regarding the protein stability of PDIP38, although, as seen in another study¹⁰, we observed no degradation of this protein in our in vitro assays (Supplementary Fig. 10a), nor did we observe changes in PDIP38 mitochondrial localization when comparing WT and CLPP^−/− HEK293 T-REx cells (Supplementary Fig. 10b,c). Conversely, a prior study reported that PDIP38 can protect and stabilize ClpX from Lon-mediated degradation in vitro¹⁰. Consistent with this, siRNA knockdown of PDIP38 led to a loss of ClpX protein in HEK293 T-REx cells (Fig. 3a), and, therefore, we propose that the increase in ClpX levels in CLPP^−/− and Perrault-related variant cells is due to a concomitant increase in PDIP38 protein. Additionally, we found that PDIP38 enhances the in vitro degradation of CHCHD2, ALAS1, and TFAM by ClpXP (Fig. 5e), a result which provides direct evidence for the role of PDIP38 as a general substrate adaptor for ClpXP.

Treatment of Perrault syndrome is currently limited to management of symptoms and related complications¹⁵. Our biochemical analysis of C144R-ClpP suggests that some Perrault-related ClpP variants retain peptidase activity but are unable to degrade folded substrates due to impaired ClpX binding (Figs. 6, 7, Supplementary Fig. 6c,d, 7). Subsequently, using the small molecule ClpP activator TR-65^12,56, we show that C144R-ClpP can be activated in vitro to degrade both α-casein and TFAM in a ClpX-independent manner (Fig. 8a,e). Interestingly, we found slightly divergent results for other ClpP substrates, where in vitro we observed degradation of CHCHD2 but not ALAS1 (Fig. 8e, Supplementary Fig. 8j), while in C144R-ClpP NPCs we observed loss of both ALAS1 and TFAM, but not of CHCHD2 (Fig. 8g). It is important to note that TR-65 does not restore the unfoldase activity observed in the intact ClpXP complex, and prior characterization of TR-27, a chemically similar ClpP activator, showed that TR-activated ClpP preferentially targets flexible regions and loops⁵⁸. Therefore, tightly folded or misfolded and/or aggregated proteins may lack the structural flexibility necessary for degradation by activated ClpP, which may be the case for CHCHD2 in C144R-ClpP cells where matrix-localized CHCHD2 is likely misfolded and aggregated. Conversely, while ALAS1 was degraded in cells, activated C144R-ClpP was unable to degrade recombinant ALAS1 in vitro, which could be due to missing post-translational modifications or missing cofactors when using bacterially-expressed ALAS1. Taken together, while TR-65 can restore proteolytic activity of C144R-ClpP to degrade folded proteins in cells, including the substrates TFAM and ALAS1, further investigation into whether ClpP activation is a valid avenue for the treatment of PRLTS3 is needed.

Overall, we have uncovered three bona fide ClpXP protein substrates and detailed their regulation by the ClpX adaptor protein PDIP38. Critically, the identified substrates may play an important role in the etiology of PRLTS3. Some of these ClpXP substrates could serve as targets for disease treatment in early stages of human development using low dosing of established ClpP activating compounds such as ONC201, which has been recently approved for the treatment of H3K27M-mutant diffuse glioma (ClinicalTrials.gov: NCT05580562)^86,87.

Methods

Ethical approval

All mammalian cell culture work was completed in accordance with and with oversight approval by institutional biosafety offices and permits. Perrault patient CLPP^T145P/T145P (c.433 A > C; 46,XX)²¹, heterozygous CLPP^T145P/WT (sibling; 46,XX), and unrelated donor control CLPP^WT/WT (46,XY) human fibroblasts were obtained for this study with written and informed consent from all participants prior to sample collection and/or study involvement. Use of patient-derived fibroblasts was performed in compliance with the World Medical Association Declaration of Helsinki (https://www.wma.net/policies-post/wma-declaration-of-helsinki/). Ethics approval for patient-derived fibroblasts was granted by the National Health Service Ethics Committee (16/WA/0017) and the University of Manchester to W.G.N.

Reagents, materials, and antibodies

Antibodies and their dilutions used during immunoblotting, immunofluorescence, and for immunoprecipitation can be found in Supplementary Data 2. Primers used for qPCR and RT-qPCR can be found in Supplementary Data 3.

Cloning of bacterial and mammalian expression constructs

Bacterial expression constructs were generated by cloning cDNA for human ClpP (residues 57-277), CHCHD2 (residues 53-151)⁸⁸, ALAS1 (residues 57-640), TFAM (residues 43-246), and PDIP38 (residues 52-368) without their mitochondrial targeting sequence (-MTS), or full-length CHCHD2 (residues 1-151), into a modified pETSUMO vector with an N-terminal His₁₂-SUMO tag (Invitrogen)⁸⁹. Human ClpX (residues 64-633) was cloned without its MTS into a pQLinkH vector (Qiagen) to express His₇-TEV-ClpX(-MTS). For mammalian expression constructs, cDNA for full-length human ClpP (+ MTS) or full-length human PDIP38 (+ MTS) were cloned into pcDNA5/FRT/TO (Invitrogen) with a C-terminal 3×FLAG tag. ClpP variants were generated by site-directed mutagenesis of the parental pETSUMO-ClpP or pcDNA5/FRT/TO-ClpP vectors according to the QuikChange II protocol (Agilent). Expression plasmids were verified by DNA sequencing.

Mammalian cell culture, transfection, and siRNA knockdown

Human HEK293 T-REx wild-type (WT) and CRISPR-knockout CLPP^−/− cells were obtained from Professor Aleksandra Trifunovic (University of Cologne, Germany)⁹⁰. Cells were cultured in Dulbecco’s modified Eagle’s medium (DMEM; Wisent) supplemented with 10% (v/v) fetal bovine serum (FBS; Corning) and 100 U/mL of Pen/Strep (Wisent). Cells were cultured at 37 °C with 5% CO₂ using standard mammalian culture practices and plasticware. Cells were passaged using trypsin-EDTA when reaching ~95% confluency. For transient expression of 3×FLAG-tagged PDIP38 and 3×FLAG-tagged ClpP, HEK293 T-REx WT and CLPP^−/− cells were transfected with pcDNA5/FRT/TO vectors at ~50% confluency using jetPRIME (Polyplus) and incubated for 18–24 h at 37 °C with 5% CO₂. Transfection media was replaced with fresh media containing 3 μg/mL doxycycline to induce heterologous protein expression and the cells were incubated for 48 h prior to collection for immunoblotting or processed for immunofluorescence. For siRNA knockdown, HEK293 T-REx cells were plated onto 12-well tissue culture plates and incubated with 100 nM of either control (MISSION #SIC001; Sigma), anti-ClpP (siGENOME SMARTpool, M-005811-01-0005; Dharmacon), or anti-PDIP38 siRNA (Silencer Select, S25057; ThermoFisher) following transfection with jetPRIME (Polyplus). Cells were incubated for 96 h at 37 °C with 5% CO₂ and collected for immunoblotting.

Patient-derived fibroblasts

Perrault-patient CLPP^T145P/T145P, heterozygous CLPP^T145P/WT, and donor-control CLPP^WT/WT human fibroblasts were cultured in high-glucose DMEM (Sigma) supplemented with 10% (v/v) FBS (Gibco) and 100 U/mL of Pen/Strep (Wisent) and grown at 37 °C with 5% CO₂ using standard mammalian culture practices and plasticware. Cells were passaged using trypsin-EDTA when reaching ~95% confluency.

Human-induced pluripotent stem cells (iPSCs) cell culture

For culturing of iPSCs, two PGP1 (Coriell Institute; GM23338, 46,XY) cell lines harboring a homozygous Perrault-related CLPP^C144R/C144R (c.403 T > C) variant (clones H1 and H3) were generated by CRISPR-editing (Synthego). Successful genome editing was confirmed by Sanger sequencing (Synthego) using the following primers: Fwd (5′−3′): CCACAGCGCCCAGTCATATT; Rev (5′−3′): AAAAACATCCCCTCCCACGC. Karyotype analysis was performed by KaryoStat+ (ThermoFisher). iPSCs were plated as small cell aggregates (clump passaging) onto standard mammalian culture plasticware in mTesR Plus media (StemCell Technologies) supplemented with 100 U/mL of Pen/Strep (Wisent). For plate coating, iMatrix-511 (Nippi) was added to the culture vessels at 0.25 μg/cm² into the cell suspension at the time of cell plating (pre-mixing method). To promote post-plating survival, cultures were treated with 10 μM Y-27632 (Cayman Chemicals), a Rho kinase inhibitor, for up to 24 h. Cells were grown at 37 °C with 5% CO₂ and refed daily with mTesR Plus media until reaching ~80-90% confluency. Cells were then dissociated from the culture vessels using ReLeSR (StemCell Technologies) and passaged as cell aggregates. For cryopreservation, iPSCs were dissociated into cell aggregates with ReLeSR, pelleted, resuspended in CryoStor CS10 (StemCell Technologies), and stored in LN₂. iPSC cultures were routinely inspected to ensure absence of spontaneous differentiation using a phase-contrast microscope, validated by RT-qPCR and by immunofluorescence for expression of pluripotency markers, and were cultured for ≤ 20 passages to minimize the potential for development of chromosomal abnormalities.

NPC differentiation and culturing

Generation of iPSC-derived neural progenitor cells (NPCs) was carried out via the dual-SMAD inhibitor approach using the STEMdiff SMADi Neural Induction Kit (StemCell Technologies) as per the manufacturer’s “monolayer culture” protocol. Briefly, iPSC cultures were grown to ~90% confluency and visually checked to ensure absence of spontaneous differentiation prior to neural induction. To begin, iPSCs were dissociated into a single-cell suspension using ReLeSR (StemCell Technologies) and 2 × 10⁶ cells from WT or each C144R-ClpP lines were subsequently pelleted, resuspended into 2 mL of STEMdiff SMADi Neural Induction media containing SMAD inhibitors and 10 μM Y-27632, and plated onto a Matrigel-coated 6-well culture plate. Matrigel coating was performed with hESC-qualified Matrigel (Corning) for 3 h at 37 °C using the lot-specified dilution factor in DMEM/F12 (Wisent). Cells were grown at 37 °C with 5% CO₂ and refed daily with STEMdiff SMADi Neural Induction media for a total of 21 days with splitting on days 7 and 14 using Accutase (Sigma) and replating of 2 × 10⁶ cells onto Matrigel-coated 6-well plates. Following neural induction, NPCs were maintained in STEMdiff Neural Progenitor Medium (StemCell Technologies), grown at 37 °C with 5% CO₂ with daily media changes until reaching confluency, and plated onto Matrigel-coated plates for immunoblot, immunofluorescence, and qPCR analysis. For cryopreservation, NPCs were dissociated with Accutase into single-cell suspensions, pelleted, resuspended in CryoStor CS10, and stored in LN₂. NPCs were validated by RT-qPCR and immunofluorescence for expression of NPC-specific markers and cultured for ≤ 10 passages to minimize further spontaneous neural differentiation.

Immunofluorescence microscopy

For immunostaining of 3×FLAG-tagged PDIP38, HEK293 T-REx cells were plated onto poly-D-lysine/laminin-coated glass coverslips [100 μg/mL poly-D-lysine (Sigma), 15 μg/mL mouse laminin (Corning), each for 2 h at 21 °C] and transfected to express PDIP38-3×FLAG as detailed above. Cells were treated with 3 μg/mL doxycycline to induce PDIP38-3×FLAG expression and the cells were incubated for 24 h until ~70% confluency. Cells were fixed with 4% paraformaldehyde in phosphate buffer saline (PBS; 137 mM NaCl, 2.7 mM KCl, 10 mM Na₂HPO₄, and 1.8 mM KH₂PO₄; pH 7.4) for 20 min at 21 °C, then permeabilized in PBS with 0.2% (v/v) Triton X-100 for 15 min followed by blocking with 5% (v/v) normal donkey serum (Sigma) in PBS for 1 h at 21 °C. Cells were subsequently incubated for ~18 h at 4 °C with primary antibodies, washed with PBS, then incubated for 1 h at 21 °C with fluorescently labelled secondary antibodies. Cells were washed with PBS, counterstained with 1 μg/mL DAPI (Sigma) in PBS for 10 min at 21 °C and mounted using ProLong Glass Antifade Mountant (ThermoFisher).

For stem cell marker analysis, iPSCs were plated as cell aggregates onto iMatrix-511-coated 35 mm plastic μ-Dishes (Ibidi), while NPCs were plated onto Matrigel-coated glass coverslips. Cells were grown at 37 °C with 5% CO₂ until ~50-80% confluent, fixed with 4% paraformaldehyde in PBS for 20 min at 21 °C, then iPSCs were concurrently permeabilized and blocked for 1 h at 21 °C in PBS with 0.2% (v/v) Triton X-100 and 5% (v/v) normal donkey serum. iPSCs were subsequently incubated for 1 h at 21 °C with primary antibodies, washed in PBS with 0.05% (v/v) Tween-20, then incubated for 1 h at 21 °C with fluorescently labelled secondary antibodies. Alternatively, NPCs were permeabilized in PBS with 0.2% (v/v) Triton X-100 for 20 min followed by blocking with 5% (v/v) normal donkey serum in PBS for 1 h at 21 °C. Immunostained cells were washed in PBS with 0.05% (v/v) Tween-20, and cells were mounted and counterstained using either Ibidi Mounting Medium with DAPI (iPSCs) or ProLong Glass Antifade Mountant with NucBlue (Hoechst 33342) Stain (NPCs; ThermoFisher). For immunofluorescent staining, antibodies were diluted in PBS with 5% (v/v) normal donkey serum.

Immunostained cells were imaged using an Eclipse 80i fluorescence microscope (Nikon) equipped with the X-Cite Series 120Q excitation light source (Excelitas Technologies) and images were captured using either a 10× air-immersion (iPSCs/NPCs) or 100× oil-immersion objective (HEK293 T-REx cells). Images were processed using the NIS-Elements Basic Research Software (Nikon).

SDS-PAGE and immunoblotting

Unless otherwise specified, cells were harvested in SDS lysis buffer [50 mM Tris-HCl; pH 8.0, 150 mM NaCl, 1% (w/v) SDS] supplemented with SIGMAFAST™ Protease (Sigma) and phosphatase inhibitor cocktails (10 mM NaF, 1 mM Na₃VO₄, 20 mM Na₂HPO₄, 3 mM β-glycerolphosphate, 5 mM sodium pyrophosphate), heated at 95 °C for 10 min, sonicated, and centrifuged at 21,000g for 10 min. Protein concentrations were determined by BCA assay (ThermoFisher) and lysates were denatured in 1×Laemmli sample buffer [62.5 mM Tris-HCl; pH 6.8, 2% (w/v) SDS, 10% (v/v) glycerol, 5% (v/v) β-mercaptoethanol, 0.002% (w/v) bromophenol blue] at 95 °C for 10 min. Protein samples were subsequently resolved on SDS-PAGE gels that were either stained with Coomassie Brilliant Blue R-250 (BioShop) or were transferred for immunoblotting onto PVDF membranes (Bio-Rad) using a modified fast transfer buffer [50 mM Tris, 20 mM HEPES, 1 mM EDTA, 10% (v/v) methanol]⁹¹ on a Trans-Blot Turbo Transfer System (Bio-Rad). For immunoblotting, membranes were blocked for 1 h at 21 °C in 5% (w/v) non-fat milk powder in PBS supplemented with 0.15% (v/v) Tween-20 (PBST) followed by overnight incubation with primary antibodies at 4 °C. Probed membranes were then washed with PBST and primary antibodies were detected using HRP-conjugated secondary antibodies (Jackson ImmunoResearch) and SuperSignal West Pico PLUS Chemiluminescent Substrate (ThermoFisher) on a ChemiDoc MP System (Bio-Rad).

Immunoblotting of patient fibroblasts

Fibroblasts were lysed on-ice in RIPA buffer [50 mM Tris–HCl; pH 8.0, 150 mM NaCl, 1% (v/v) IGEPAL CA-630, 0.5% (w/v) sodium deoxycholate, 0.1% (w/v) SDS] supplemented with protease inhibitor cocktail (Promega). Whole-cell lysates were centrifuged at 21,000g for 10 min, protein concentrations were determined by BCA assay, and lysates were denatured in 1×Laemmli sample buffer with 100 mM DTT at 95 °C for 5 min. Protein samples were subsequently resolved on SDS-PAGE gels and transferred onto PVDF membranes (Sigma) using a Trans-Blot Semi-Dry Transfer Cell System (Bio-Rad). Membranes were washed with 1×Tris-buffered saline (TBS; 20 mM Tris-HCl; pH 7.6, 150 mM NaCl) supplemented with 0.1% (v/v) Tween-20 (TBS-T) and blocked with 5% (w/v) non-fat milk powder in TBS-T. For immunoblotting, membranes were incubated overnight at 4 °C with primary antibodies, washed with TBS-T, then incubated with IRDye secondary antibodies (LICORbio) for 1.5 h at 21 °C. Finally, membranes were washed with TBS-T and visualized with an Odyssey Fc Imager (LICORbio) system.

Depletion of mitochondrial DNA (mtDNA)

HEK293 T-REx cells, iPSCs, and NPCs were plated onto 12-well tissue culture plates at a density of 1.2 × 10⁵, 1.0 × 10⁵, or 6.5 × 10⁵ cells/well, respectively, and were incubated with 25–500 μM 2′,3′-dideoxycytidine (ddC; Sigma) for 96 h at 37 °C with 5% CO₂. Control cells were vehicle-treated (H₂O). For patient-derived fibroblasts, cells were plated onto 6-well plates (mtDNA/mRNA) or T75 flasks (immunoblotting) at a density of 2.0 × 10⁴ cells/cm² and treated with 100 μM ddC for 96 h at 37 °C with 5% CO₂ in DMEM supplemented with 50 μg/mL uridine (Sigma). Cells were harvested and immunoblotted as described above or processed for assessment of mtDNA depletion by qPCR for the mitochondrial genes ATP6, ND1, and CO1 following DNA extraction with either the PureLink Genomic DNA Mini Kit (HEK293 T-REx cells, iPSC/NPCs; ThermoFisher) or the QIAamp Fast DNA Tissue Kit (Fibroblasts; Qiagen). As a control, TFAM mRNA levels following ddC treatment were measured separately by RT-qPCR (see next section).

RT-qPCR

RNA from HEK293 T-REx cells, iPSCs, and NPCs was isolated using the PureLink RNA Mini Kit (ThermoFisher) with on-column DNA digestion and was subsequently converted to cDNA using GoScript Reverse Transcriptase (Promega). qPCR for mtDNA and cDNA targets was performed on a QuantStudio 3 Real-Time PCR System (ThermoFisher) using Luna Universal qPCR Master Mix (NEB) as per the manufacturer’s guidelines. Alternatively, RNA from patient-derived fibroblasts was isolated using the RNeasy Mini Kit (Qiagen) with on-column DNA digestion and was subsequently converted to cDNA using GoScript Reverse Transcriptase. qPCR for mtDNA and cDNA targets from fibroblast samples was performed on a StepOne Real-Time PCR System (ThermoFisher) using PowerUP SYBR Green Master Mix (ThermoFisher) as per the manufacturer’s guidelines. For a list of qPCR primers, see Supplementary Data 3. Data were analyzed using QuantStudio Design & Analysis Software (ThermoFisher) or StepOne Software (ThermoFisher) and quantified according to the ΔΔCT method⁹².

Cycloheximide chase assay

To measure cellular turnover of ClpXP substrates, HEK293 T-REx or iPSCs/NPCs were treated with 100 μg/mL or 200 μg/mL of cycloheximide (Sigma), respectively, for 1–24 h and collected at desired time points. Control cells were DMSO-treated. For the determination of TFAM protein half-life following depletion of mtDNA, cells were first treated with ddC for 96 h prior to treatment with cycloheximide. Cells were lysed in SDS lysis buffer and assessed by immunoblotting. To determine protein half-life, immunoreactive bands were quantified by densitometry, plotted on a semi-log scale, and fitted with a linear regression model to calculate the slope of decay (α). Protein half-life, T_1/2, was calculated according to first order kinetics where T_1/2 = ln(2)/α⁹³.

Triton X-100 fractionation of cells

Protein solubility of endogenous CHCHD2, ALAS1, and TFAM, or of transiently expressed ClpP-3×FLAG, was assessed in either WT or CLPP^−/− HEK293 T-REx cells by lysing cells on ice in Triton lysis buffer [10 mM HEPES; pH 7.4, 5 mM MgCl₂, 1 mM EGTA, 150 mM NaCl, 5% (v/v) glycerol, 1% (v/v) Triton X-100] supplemented with protease/phosphatase inhibitors. Lysates were centrifuged at 4 °C for 15 min at 20,000g, protein concentration was determined by BCA assay, and aliquots from the Triton-soluble supernatant were prepared for immunoblotting by denaturing in 1×Laemmli sample buffer. Triton-insoluble proteins were prepared by denaturing the pellet in equal volumes of 2×Laemmli sample buffer at 85 °C for 15 min with intermittent vortexing prior to separation on SDS-PAGE gels. As controls, total lysates were prepared separately by lysing cells in SDS lysis buffer prior to immunoblotting.

Crude mitochondria isolation and submitochondrial fractionation

To isolate crude mitochondria, 3 × 10⁷ of WT or CLPP^−/− HEK293 T-REx cells were harvested, washed on-ice in PBS, and lysed with a 27-gauge syringe (10 passages) in mitochondria isolation buffer (5 mM HEPES; pH 7.4, 220 mM mannitol, 70 mM sucrose, 1 mM EGTA) supplemented with protease and phosphatase inhibitors. The homogenate was centrifuged at 700 g for 10 min at 4 °C to remove cell debris after which the lysate was cleared an additional time at 1200 g for 10 min at 4 °C. Mitochondria were then separated from the cytosolic supernatant by centrifugation at 10,000 g for 10 min at 4 °C, and the pellet was washed once with mitochondria isolation buffer to remove contaminating cytosolic proteins. For immunoblotting, the pellet was lysed in SDS lysis buffer with protease/phosphatase inhibitors and protein concentration was determined by BCA assay. Subsequently, protein samples from both the mitochondrial pellet and cytosolic supernatant were denatured in 1×Laemmli sample buffer and separated on SDS-PAGE gels. As controls, total cell lysates were prepared separately by lysing cells in SDS lysis buffer prior to immunoblotting.

Submitochondrial fractionation was completed as described previously⁷³ with minor adjustments. Briefly, pelleted mitochondria obtained as above were resuspended in PBS with 250 mM sucrose (Isotonic buffer) and protein concentration was determined by BCA assay. 80 µg samples of crude mitochondria were then transferred to separate 1.5 mL microcentrifuge tubes, centrifuged at 10,000g for 5 min at 4 °C, and gently resuspended in either Isotonic, Hypotonic (10 mM HEPES, pH 7.4) or Triton lysis buffer (above). Mitochondria samples were subsequently incubated on-ice for 90 min with 8 µg of Proteinase K (Invitrogen) to digest outer mitochondrial membrane (OMM) proteins, inner mitochondrial membrane (IMM) proteins, or total mitochondrial proteins in Isotonic, Hypotonic, or Triton-lysed samples, respectively. Disappearance of the immunoblotted proteins following PK digestion indicates the location of these proteins within the mitochondria. The reaction was terminated with the addition of 2 mM PMSF, after which the mitochondria were denatured in 1×Laemmli buffer and 15 μg of protein from each sample were separated on SDS-PAGE gels and assessed by immunoblotting.

Subcellular fractionation

To fractionate cells into nuclear, cytosolic, and mitochondrial fractions, 3 × 10⁷ of WT or CLPP^−/− HEK293 T-REx cells were harvested, washed on-ice in PBS, and lysed using the Mitochondria Isolation Kit for Cultured Cells (ThermoFisher). After lysis, cells were pelleted at 700g for 10 min (nuclear fraction) at 4 °C and the resulting supernatant was then centrifuged at 12,000g for 15 min at 4 °C to isolate cytosolic (supernatant) and mitochondrial (pellet) fractions. The nuclear and mitochondrial fractions were subsequently washed three-times each with PBS supplemented with either 0.5% (v/v) Triton X-100 or 250 mM sucrose, respectively. Once isolated, the nuclear, cytosolic, and mitochondrial fractions were denatured in 1×Laemmli buffer and 7 μg of protein from each sample were separated on SDS-PAGE gels and assessed by immunoblotting.

Co-immunoprecipitation (co-IP)

For co-IP of endogenous ClpP, ClpX, ALAS1, CHCHD2, and TFAM with PDIP38, HEK293 T-REx cells were lysed on-ice in Triton lysis buffer supplemented with protease/phosphatase inhibitors. Lysates were centrifuged, protein concentration was determined by BCA assay, and the clarified lysates were incubated overnight with α-PDIP38 primary antibody (Proteintech; 15080-1-AP) at 4 °C in a ratio of 1 μg antibody to 1 mg lysate. α-PDIP38 immune complexes were captured onto protein-G Dynabeads (ThermoFisher) for 1 h at 4 °C, washed with Triton lysis buffer, then eluted into 2×Laemmli sample buffer at 85 °C for 10 min. Alternatively, co-IP of endogenous ClpX-ClpP from iPSC-derived NPCs was carried out as described above except using α-ClpP primary antibody (Proteintech; 15698-1-AP). For co-IP of endogenous ClpX with overexpressed ClpP, 3×FLAG-tagged ClpP was transiently expressed in HEK293 T-REx CLPP^−/− cells and was captured onto anti-FLAG M2 Magnetic Beads (Sigma) following lysis with Triton lysis buffer. α-FLAG immune complexes were washed with Triton lysis buffer and were eluted into 2×Laemmli sample buffer at 85 °C for 10 min. Co-IP samples were subsequently analyzed by immunoblotting. As a control, total protein samples were prepared separately by lysing cells in SDS lysis buffer prior to immunoblotting.

Mass spectrometry from patient-derived fibroblasts

Proteomic data was acquired by performing SILAC-proteomics on patient-derived CLPP^T145P/T145P (c.433 A > C), heterozygous CLPP^T145P/WT, and donor-control CLPP^WT/WT fibroblasts as part of a larger proteomics dataset to be released separately in the future. For this study, whole-proteome steady-state abundance was analyzed without SILAC modifications during data processing. Primary fibroblasts were plated onto 60 mm plates and cultured for 12 days in SILAC labeling medium supplemented with ‘heavy’ ¹³C- and ¹⁵N-containing arginine and lysine (ThermoFisher). Subsequently, SILAC media was removed, and cells were cultured in DMEM containing ‘light’ ¹²C- and ¹⁴N-isotopes (ThermoFisher) for 9 h. Cells were harvested into ice-cold PBS, pelleted, and lysed in 50 mM triethylammonium bicarbonate buffer, pH 7.55, supplemented with 5% (w/v) SDS, cOmplete protease inhibitors (Roche), and Pierce Universal Nuclease (ThermoFisher). Lysates were centrifuged at 16,000 g for 20 min and protein concentration of the clarified supernatant was determined by BCA assay. 10 μg aliquots of protein were subsequently reduced with the addition of 20 mM TCEP for 30 min at 47 °C, then alkylated with 10 mM iodoacetamide for 20 min at 21 °C with protection from light. Protein clean-up and preparation of trypsin-digested peptides prior to liquid chromatography mass spectrometry (LC-MS) was performed on-column using S-Trap columns (ProtiFi) according to the manufacturers protocol. Digested peptides were dried using a SpeedVac concentrator (ThermoFisher), then resuspended in 30 µL of LC-MS Buffer A [0.1 % (v/v) formic acid in HPLC-grade water].

LC-MS was performed on an Evosep One system (Evosep Biosystems) with a 15 cm Aurora Elite C18 column with integrated captive spray emitter (IonOpticks) and an in-line timsTOF HT mass spectrometer (Bruker) operated in dia-PASEF mode. Briefly, ~500 ng of peptides resuspended in LC-MS Buffer A were loaded onto desalting C18 EvoTips (Evosep Biosystems) and separated at 50 °C over a 0-35% gradient of LC-MS Buffer B [0.1 % (v/v) formic acid in acetonitrile] using a predefined Evosep One Whisper-Zoom 20 SPD protocol (100 nL/min or 200 nL/min, 58 min run-time). Mass and IM ranges were set to 300–1200 m/z and 0.6–1.45 1/K₀ respectively, and dia-PASEF was performed as previously described⁹⁴ using variable width IM-m/z windows without overlap. TIMS ramp and accumulation times were 66 ms, total cycle time was ~1.8 s. Collision energy was applied in a linear fashion where ion mobility = 0.6-1.6 1/K₀ and collision energy = 20–59 eV.

For data analysis, raw data files were initially searched using the publicly available DIA-NN software package (1.8.1) against the H. sapiens proteome (UniProt, 42,519 entries; accessed 15-07-2024) with the following peptide mass modifications: fixed carbamidomethyl (C) and variable oxidation (M). Mass accuracy and MS1 accuracy were set to 15, peptide length range = 7–30, precursor charge range = 1–4, precursor m/z range = 300–1800, and fragment ion m/z range = 200–1800. Report files were then processed using RStudio (Posit Software) to separate channels followed by protein quantification using the iq package⁹⁵ to calculate the MaxLFQ based on precursor translated intensities. Following contaminant removal, protein intensities were log₂-transformed and proteins identified by only one unique peptide were removed from the analysis to increase data stringency. Normalization was applied based on median intensities and the fold-change between groups was determined using the Benjamini-Hochberg correction method within the limma R/Bioconductor software package⁹⁶. Volcano plot comparisons between different fibroblast samples were plotted in GraphPad Prism according to p-value and log₂(fold change). Lastly, gene set enrichment analysis (GSEA) was performed for gene ontologies (GOs) and Kyoto Encyclopedia of Genes and Genomes (KEGG) pathways using the ClusterProfiler package⁹⁷ and plotted according to q-value using SPR plot⁹⁸. Mass spectrometry data has been deposited in the PRotermics IDEntifications database (PRIDE) with accession number PXD070104.

Recombinant protein expression and purification

His₁₂-SUMO-ClpP(-MTS) plasmids were transformed into clpP-deleted SG1146 BL21(DE3) E. coli (ΔclpP::cat) to prevent the co-purification of E. coli ClpP with recombinant human ClpP. His₁₂-SUMO-CHCHD2(+ MTS), His₁₂-SUMO-CHCHD2(-MTS), His₁₂-SUMO-ALAS1(-MTS), His₁₂-SUMO-TFAM(-MTS), and His₁₂-SUMO-PDIP38(-MTS) plasmids were transformed into E. coli BL21(DE3) pRIL. His₇-TEV-ClpX(-MTS) plasmid was transformed into E. coli BL21(DE3) ΔclpP::cat ΔclpX::kanR to prevent the co-purification of E. coli ClpXP. Recombinant protein purification was performed as follows:

ClpP, CHCHD2, TFAM, ALAS1, and PDIP38 - Transformed E. coli cells were grown at 37 °C in LB broth (10 g/L tryptone, 5 g/L yeast extract, 10 g/L NaCl) supplemented with 30 μg/mL kanamycin until OD₆₀₀ ~ 0.6. Recombinant protein expression was induced by addition of 1 mM isopropyl β-D-thiogalactopyranoside (IPTG) and cultures were grown at 18 °C for 16–18 h (ClpP, CHCHD2, and PDIP38) or 37 °C for 4 h (TFAM). For recombinant ALAS1 and T145P-ClpP, transformed cells were grown in auto-induction media [16 g/L tryptone, 10 g/L yeast extract, 50 mM Na₂HPO₄, 50 mM KH₂PO₄, 25 mM (NH₄)₂SO₄, 2 mM MgSO₄, 0.5% (v/v) glycerol, 0.5 g/L glucose, 2 g/L lactose]⁹⁹ supplemented with 300 μg/mL kanamycin at 16 °C for 72 h. Cells were harvested by centrifugation, resuspended in buffer containing 25 mM Tris-HCl; pH 7.5, 500 mM NaCl, 10 mM imidazole, and 10% (v/v) glycerol, supplemented with lysozyme, and lysed by sonication. For ALAS1, 0.5 mM TCEP (Sigma) was added throughout the purification process. Cell debris was removed by centrifugation at 35,000g for 30 min at 4 °C and the clarified supernatant was applied to a gravity column loaded with HisPur Ni-NTA resin (ThermoFisher) to purify His-tagged proteins by Ni²⁺ affinity purification using standard procedures. The eluates for ClpP, CHCHD2, TFAM, and PDIP38 were subsequently dialyzed overnight at 4 °C into buffer [25 mM Tris-HCl; pH 7.5, 200 mM NaCl, 10% (v/v) glycerol]. Alternatively, ALAS1 was dialyzed overnight at 4 °C into buffer [25 mM Tris-HCl; pH 7.2, 500 mM NaCl, 0.5 mM TCEP, 10% (v/v) glycerol]. To remove the N-terminal His₁₂-SUMO tag, Ni²⁺-purified proteins were incubated overnight with 50 μg/mL recombinant His₆-Ulp1 SUMO protease (purified in-house) during dialysis. Untagged ClpP, CHCHD2, TFAM, ALAS1, and PDIP38 were further purified by loading onto HisPur Ni-NTA resin to capture His₆-Ulp1 and collecting the flow-through. Finally, purified proteins were concentrated, aliquoted, flash-frozen in LN₂, and stored at −80 °C. Protein concentrations were determined using a Nanodrop spectrophotometer (ThermoFisher) at 280 nm using the extinction coefficient values for each protein as determined using ProtParam (ExPASy).

ClpX – Transformed E. coli cells were grown at 37 °C in Terrific Broth [12 g/L tryptone, 24 g/L yeast extract, 0.5% (v/v) glycerol, 17 mM KH₂PO₄, 72 mM K₂HPO₄] supplemented with 100 μg/mL ampicillin until OD₆₀₀ ~ 0.6. Recombinant ClpX expression was induced by addition of 1 mM IPTG and cultures were grown at 18 °C for 16–18 h. Cells were harvested by centrifugation, resuspended in buffer containing 25 mM Tris-HCl; pH 7.5, 500 mM KCl, 5 mM MgCl₂, 0.5 mM TCEP, and 10% (v/v) glycerol, supplemented with lysozyme, and lysed by sonication. Cell debris was removed by centrifugation at 35,000 g for 30 min at 4 °C and the clarified supernatant was applied to a gravity column loaded with HisPur Ni-NTA resin to purify His₇-tagged ClpX by Ni²⁺ affinity purification using standard procedures. The eluate was subsequently dialyzed overnight at 4 °C into TEV cleavage buffer [25 mM Tris-HCl; pH 8.0, 100 mM KCl, 5 mM MgCl₂, 0.5 mM EDTA, 0.5 mM TCEP, 10% (v/v) glycerol]. To remove the N-terminal His₇-TEV tag, Ni²⁺-purified His₇-ClpX was incubated overnight with 50 μg/mL recombinant His₆-TEV protease (purified in-house) during dialysis. Untagged ClpX was then briefly dialyzed into storage buffer [25 mM Tris-HCl; pH 8.0, 100 mM KCl, 5 mM MgCl₂, 0.5 mM TCEP, 10% (v/v) glycerol] and subsequently purified by loading onto HisPur Ni-NTA resin to capture His₆-TEV and collecting the flow-through. Finally, purified ClpX was concentrated, aliquoted, flash-frozen in LN₂, and stored at −80 °C. Protein concentration was determined as outlined above using a Nanodrop spectrophotometer.

Tag-intact PDIP38 and ClpX – His₁₂-SUMO-tagged PDIP38 and His₇-TEV tagged ClpX were used as bait proteins for in vitro pulldowns. For purification of tag-intact His₁₂-SUMO-PDIP38, after elution from the Ni-NTA column as described above, the eluate was instead dialyzed overnight at 4 °C into ion exchange buffer [25 mM Tris-HCl; pH 8.0, 50 mM KCl, 70 mM L-Arginine, 30 mM L-Glutamic acid, 10% (v/v) glycerol]. The eluate was loaded onto a MonoQ 5/50 GL column (Cytiva), and proteins were separated by anion exchange chromatography over a 50–500 mM KCl gradient in ion exchange buffer using an AKTA Purifier FPLC system (Cytiva). Protein samples following anion exchange were run on SDS-PAGE gels and fractions containing purified His₁₂-SUMO-PDIP38 were pooled. To purify tag-intact His₇-TEV-ClpX, after elution from the Ni-NTA column, protein samples were run on SDS-PAGE gels and fractions containing purified His₇-TEV-ClpX were pooled and dialyzed overnight at 4 °C into ClpX storage buffer (above). Purified His₁₂-SUMO-PDIP38 and His₇-TEV-ClpX were concentrated, aliquoted, flash-frozen in LN₂, and stored at −80 °C. Protein concentration was determined as above using a NanoDrop spectrophotometer.

Size-exclusion chromatography

200 μg of purified recombinant WT or Perrault-variant ClpP were loaded onto a Superdex 200 Increase 10/300 GL column (Cytiva) and were separated into 0.5 mL fractions in running buffer [25 mM Tris-HCl; pH 7.5, 300 mM NaCl, 10% (v/v) glycerol] at 0.5 mL/min using an AKTA Purifier FPLC system (Cytiva) at 4 °C. Elution of ClpP from the column was monitored at 280 nm, and protein samples from each fraction were denatured in 1×Laemmli sample buffer, run on SDS-PAGE gels, and stained with Coomassie Brilliant Blue R-250. Chromatographic traces were prepared in GraphPad Prism. For column calibration, the following proteins were used: thyroglobulin (669 kDa; Sigma), apoferritin (443 kDa; Sigma), β-amylase (200 kDa; Sigma), alcohol dehydrogenase (150 kDa; Sigma), and bovine serum albumin (66 kDa; Sigma). Void volume was determined using Blue Dextran (Sigma).

In vitro pulldowns

To demonstrate direct in vitro interactions, purified His₁₂-SUMO-PDIP38 or His₇-TEV-ClpX was incubated with recombinant CHCHD2, ALAS1, or TFAM, or with PDIP38, respectively, in PDIP38 pulldown buffer (20 mM Tris-HCl; pH 7.5, 250 mM NaCl, 0.5 mM TCEP, 25 mM imidazole) for 2 h at 4 °C. For in vitro interaction of mutant ClpP with ClpX, purified His₇-TEV-ClpX was incubated with recombinant ClpP in ClpX pulldown buffer [25 mM HEPES; pH 7.5, 300 mM KCl, 5 mM MgCl₂, 5% (v/v) glycerol, 0.5 mM TCEP, 40 mM imidazole, 1 mM ATP] for 1 h at 37 °C. Complexes were captured onto Ni-NTA agarose beads (ThermoFisher), washed, and eluted with either PDIP38 or ClpX pulldown buffer, respectively, supplemented with 400 mM imidazole. Eluted protein samples were denatured in 1×Laemmli sample buffer and analyzed by SDS-PAGE and immunoblotting.

Peptidase and FITC-casein proteolytic assays

The in vitro peptidase activity of purified recombinant ClpP was determined using Trp-Leu-Ala-7-amido-4-methylcoumarin (WLA-AMC; R&D Systems). Briefly, WT or Perrault-variant ClpP (2 μM protomer) were incubated with 75 μM WLA-AMC in the absence or presence of recombinant ClpX (2 μM protomer). Alternatively, the activity of activated ClpP was determined by incubating purified ClpP (4 μM protomer) with varying concentrations (0.39–12.5 μM) of fluorescein isothiocyanate (FITC)-labeled casein (Sigma) in the presence of 4 μM TR-65 (Madera Therapeutics). All reactions were carried out at 37 °C in reaction buffer [50 mM Tris-HCl; pH 8.0, 100 mM KCl, 10 mM MgCl₂, 5% (v/v) glycerol, 0.02% (v/v) Triton X-100, 1 mM DTT] on a 96-well black flat-bottom plate (Greiner). ClpP peptidase reactions where ClpX was included were supplemented with 5 mM ATP for ClpX activity. Fluorescence was monitored at an Ex/Em of 360/440 nm (WLA-AMC) or 485/535 nm (FITC-casein) on a CLARIOstar Plus plate reader (BMG Labtech). The instrument was set to measure either every 60 s (WLA-AMC) or 30–70 s (FITC-casein) with 20 flashes per read.

To calculate V_max, K_M, and k_cat, initial rates were plotted versus FITC-casein concentration, and the data were fitted to the Michaelis-Menten equation where V₀ is the initial rate, V_max is the maximal reaction velocity, [ClpP] is the ClpP protomer concentration, and [S] is the substrate concentration. Subsequently, k_cat was calculated from V_max as follows:

$$V_0=\frac{V_{\max }\left[S\right]}{K_M+\left[S\right]}$$ 1

$$k_{cat}=\frac{V_{\max }}{\left[ClpP\right]}$$ 2

To calculate the apparent binding constant (K_d,app) of TR-65 or ADEP-28 to ClpP, raw data were fit to a linear regression model to obtain initial rates from titrating varying concentrations of ADEP-28 with 2 µM ClpP or titrating both TR-65 and ClpP with 5 μM FITC-casein. Data were processed according to the Hill equation where V₀ is the initial rate, V_max is the maximal reaction velocity, and h is the Hill coefficient:

$$V_0=\frac{V_{\max }{\left[TR-65orADEP-28\right]}^h}{{{EC}_{50}}^h+{\left[TR-65orADEP-28\right]}^h}$$ 3

For TR-65, EC₅₀ values were subsequently plotted versus ClpP concentration, and a linear regression model was fitted to obtain K_d,app, where n is the number of TR-65 binding sites per protomer of ClpP:

$${EC}_{50}=K_{d,app}+\frac{n\left[ClpP\right]}{2}$$ 4

In vitro ClpXP substrate degradation assays

Degradation of bovine α-casein (4 µM; Sigma) or of purified recombinant ALAS1 (2 µM), TFAM (5 µM), CHCHD2(+ MTS) (15 µM), and CHCHD2(-MTS) (15 µM) was carried out at 37 °C in the presence or absence of ClpP (4 µM protomer), ClpX (12 µM protomer), PDIP38 (4 µM), or TR-65 (4 µM) in PD buffer [25 mM HEPES; pH 7.5, 35 mM KCl, 25 mM MgCl₂, 0.03% (v/v) Tween-20, 10% (v/v) glycerol, 16 mM creatine phosphate, 4 mM ATP, 1 mM DTT, 13U of creatine kinase (Sigma)]. Reactions were terminated by addition of 1×Laemmli buffer, and protein samples were denatured at 95 °C for 5 min and analyzed on Coomassie Brilliant Blue (CBB)-stained SDS-PAGE gels. Alternatively, in assays that combined recombinant ALAS1 (-MTS; 64.4 kDa) and ClpX (-MTS; 62.7 kDa), proteins were transferred to PVDF and immunoblotting was performed. To test the in vitro degradation of DNA-bound TFAM, 5 µM of TFAM was preincubated for 10 min at 21 °C with 10 µM of double-stranded oligonucleotide corresponding to the TFAM binding site upstream of the mtDNA light strand promoter (uLSP)^44,55 or a scrambled sequence. Subsequently, ClpP, ClpX, and PDIP38 were added to the reaction as described above and incubated at 37 °C for 1–3 h. As controls, oligonucleotide supplemented reactions were prepared with 4 µM of either α-casein or 15 µM CHCHD2.

Crystallization and structure solution of C144R-ClpP and its complex with TR-65

In vitro purified C144R-ClpP was concentrated to 10-15 mg/mL using an Amicon Ultra-15 centrifugal filter unit (10 kDa MWCO; Sigma) and subsequently incubated at 37 °C for 1 h with 1 mM TR-65 to form TR-65-ClpP complexes. Protein crystals of apo- or TR-65-bound C144R-ClpP were grown at 23 °C for ~14 days using the sitting drop vapor diffusion method containing drops of a 1:1 mixture of protein-drug complexes with a crystallization solution containing 100 mM sodium acetate trihydrate, pH 4.6, and 4% (w/v) polyethylene glycol 4000. Once grown, crystals were harvested by first briefly soaking in crystallization solution supplemented with 20% (v/v) glycerol, then frozen immediately in liquid nitrogen. Diffraction data were collected using synchrotron X-rays at the Canadian Light Source CMCF-BM (08B1) bending magnet beamline and were reduced using the HKL2000 software¹⁰⁰. The crystal structures were solved by molecular replacement in Phenix Phaser¹⁰¹ using the protein-only model of a human ClpP monomer as a search model (PDB: 6BBA). C144R-ClpP structures were initially refined in Phenix¹⁰¹ with simulated annealing and coordinate shaking to remove model bias. Subsequent model building and refinement were performed in Coot¹⁰² and Phenix with optimization of individual coordinates, B-factors, and occupancy parameters. Protein-bound TR-65 molecules were confirmed by sharp differences in electron density and were modeled using Phenix eLBOW¹⁰³ with occupancies equal to 1.00. Data collection and refinement statistics are summarized in Table 1, and solved structures were deposited in the RCSB Protein Data Bank (PDB).

Statistical analysis

For immunoblots, densitometry from immunoreactive bands was quantified using ImageLab software (Bio-Rad) and normalized to tubulin, GAPDH, or β-actin levels. For Coomassie gels, band intensities were quantified using ImageJ software (Fiji). Data were graphed as mean ± standard deviation (SD) from independent biological replicate experiments (n) using GraphPad Prism (8.1.0). Statistical analysis was performed in GraphPad Prism using the following statistical tests: Student’s two-tailed t-test, one-way ANOVA with Bonferroni’s post hoc test, or by two-way ANOVA with Bonferroni’s post hoc test. Statistical significance was set to p ≤ 0.05.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

Conceptualization: M.E.A., T.M.M., and W.A.H. Methodology: M.E.A., T.M.M., W.A.H., M.F.M., J.J.M., H.B.T., B.B.A.R., M.T., and U.R. Investigation: M.E.A., T.M.M., M.F.M., J.J.M., H.B.T., B.B.A.R., M.T., J.Y., A.T., and K.L.L. Data analysis: M.E.A., T.M.M., M.F.M., J.J.M., H.B.T., B.B.A.R., and M.T. Manuscript preparation: M.E.A., T.M.M., and W.A.H. All authors: manuscript review and editing. Supervision: M.T., R.W.T., R.T.O., U.R., W.G.N., and W.A.H. Funding acquisition: M.T., R.W.T., R.T.O., U.R., W.G.N., and W.A.H.

Peer review

Peer review information

Nature Communications thanks David Dougan, Luca Scorrano and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. A peer review file is available.

Funding

This work was supported by grants from CIHR (PJT-173345) to WAH, the Sigrid Jusélius Foundation (Grant nr 8123) and the Newcastle University Academic Track (NUAcT) Fellowship Program to UR, NIHR Manchester Biomedical Research Centre (NIHR203308) to WGN, and by the Medical Research Council (MR/W019027/1) to RWT, RTO, and WGN. RWT is also funded by the Wellcome Centre for Mitochondrial Research (203105/Z/16/Z), the Lily Foundation, the UK NIHR Biomedical Research Centre for Ageing and Age-related disease award to the Newcastle upon Tyne Foundation Hospitals NHS Trust, the LifeArc Centre for Rare Mitochondrial Diseases (Reference: 10748), MDUK (23SI-PRG60-0013) and the UK NHS Highly Specialised Service for Rare Mitochondrial Disorders of Adults and Children.

Data availability

Mass spectrometry data has been deposited in the PRotermics IDEntifications (PRIDE) database with accession number PXD070104. Crystal structure data generated in this study have been deposited in the Protein Data Bank (PDB) database under accession codes 9PN7 and 9PNC. Reference structures that were used in this publication are deposited under the following PDB accession codes: 1TG6, 9DKV, 9YKZ, 9YKX, and 7UVR. All data generated and analyzed during this study are included in this published article and/or corresponding Supplementary Information files. Source data are provided with this paper.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at https://doi.org/10.1038/s41467-026-77888-0.