This study is compliant with the guidelines of the Declaration of Helsinki and under protocols approved by the local ethics committees. Informed consent to use these samples for research was obtained in an anonymized way. PITRM1^R183Q cell line (here defined PT1) carrying homozygous missense mutations (c.548G>A, p.Arg183Gln) was provided by the University of Bergen (Norway) and previously described by us (Brunetti et al., 2016); PITRM1^V747Gfs*31 cell line (here defined PT2) carrying homozygous c.2239dupG variant in PITRM1 (p.Val747Glyfs*31) was provided by the IRCCS Fondazione Stella Maris (Pisa, Italy) and previously described by us (Tolomeo et al., 2021). PITRM1^T931M cell line (here defined PT3) carrying homozygous PITRM1 mutation (c.2795C>T, p.Thr931Met) was provided by the Sharee Zedek Medical Center (Jerusalem, Israel) and previously described by us (Langer et al., 2018). See Table 1 for genotype-phenotype details. Controls cells were obtained by healthy donors at the Fondazione IRCCS Istituto Neurologico Carlo Besta (Milan, Italy).

Skin fibroblasts from healthy subjects (CTRL M, Z and B) and patients PT1, PT2 and PT3 were cultured in Dulbecco’s modified Eagle’s medium (DMEM) with high glucose (Euro Clone, ECB7501L), supplemented with 10% fetal bovine serum (FBS, Gibco, 100-18B-1MG), 1% penicillin/streptomycin (Gibco, ECB3001D), 1% L-glutamine (Gibco, ECB3000D), 10 μg/mL of Fibroblast grow factor (FGF, Peprotech, 100-18B-1 MG) and kept at 37°C in a 5% CO₂ incubator. Pioglitazone (PGZ, Sigma-Aldrich, CDS021593-50MG) was dissolved in dimethyl sulfoxide (DMSO, Sigma-Aldrich, D2650-100ML) to a stock solution of 10 mM and added to the cell growth medium at the final concentration of 5 µM for 48 h. As untreated condition, cells were cultured with an equal volume of DMSO in the growth medium. The medium was exchanged every day with fresh medium. In some experiments, cells were treated with Insulin at the final concentration of 1 µM for 5 days and H₂O₂ at the final concentration of 800 µM for 48 h.

Measurement of alterations in the mitochondrial membrane potential (ΔΨm) was carried out by utilizing the JC1 staining kit (Sigma, CS0390) following manufacturer’s instructions. In cells with high ΔΨm, JC-1 molecules assemble into complexes referred to as J-aggregates, yielding a red to orange fluorescence (λ_em = 590 ± 17.5 nm), conversely, cells with low ΔΨm retain JC-1 in its monomeric form, resulting in solely green fluorescence (λ_em = 530 ± 15 nm).

Fibroblasts were cultured in glass-bottom dishes and incubated with JC1 in a 37°C, 5%CO₂ incubator for 25 minutes, then washed twice with fresh PBS and incubated with phenol red-free basal medium for live imaging data acquisition.

Mitochondrial morphology was analysed on TMRM-stained cells by using ImageJ software. Fibroblasts were cultured in glass-bottom dishes using fibroblast growth medium. Subsequently, the cells were washed with PBS and incubated at 37°C for 30 min in medium supplemented with a 20 nM TMRM (Invitrogen) and an additional 5 µM Hoechst dye (Invitrogen) to simultaneously stain the nuclei. Following the incubation period, the medium was removed, and the cells were washed with PBS. Images were captured using fluorescent confocal microscopy. All confocal images were obtained using a Leica SP8 microscope with identical acquisition parameters, specifically at an emission wavelength of 574 nm.

In brief, RGB images from fluorescent microscopy were converted into binary (black and white) images to define objects and background (segmentation), followed by filtering or preprocessing including noise reduction, background subtraction, contrast and feature enhancement filters. Particle Analysis was performed on selected region-of-interest (ROI), blanking the area outside of the selection, and thresholding. Morphological measures for each ROI were annotated (Merrill et al., 2017).

Total mRNA was isolated using an RNA isolation kit (Qiagen, Germany). Following the reverse transcription reaction with the Goscript Reverse Transcriptase kit (Promega, United States, A5001), a quantitative SYBR Green-based PCR reaction was performed using iTaq universal SYBR Green Supermix (Biorad, United States, 1725124) and the following conditions: 95°C for 2 min; 95°C for 15 s; 60°C for 1 min; 60°C for 31 s; 65°C for 5 s + 0.5°C/cycle, 39 cycles. The reaction was monitored with a CFX96 Real-Time PCR system (Biorad, United States). The expression level of each gene was normalized to the housekeeping gene ACTB encoding for b-actin. Fold changes in gene expression were calculated using the 2^−ΔΔCT method, based on biological reference samples and housekeeping gene for normalization. Primers are listed in Table 2.

Proteins were extracted using an ice-cold 1X RIPA buffer supplemented with protease and phosphatase inhibitors from Roche (Switzerland). The extraction process involved centrifugation at 14,000 rpm and 4°C for 30 min, followed by three freeze-thaw cycles. The protein concentration in the resulting supernatant was determined using the Bradford protein assay kit from Biorad (CA, United States, 5000006). For gel electrophoresis, 30 μg of the protein lysate were loaded onto a polyacrylamide gel with a density ranging from 4% to 12%. The proteins were then transferred onto a nitrocellulose membrane. To block the blots, a solution of 5% milk powder or 5% BSA in TBS with 0.1% Tween-20 (TBST) was used. The membranes were incubated overnight at 4°C with primary antibodies diluted in either milk or BSA blocking solution. Subsequently, the membranes were incubated with corresponding HRP-conjugated secondary antibodies from Sigma-Aldrich (MO, United States) for 1 h at room temperature. Visualization of the proteins was achieved using either Clarity Western ECL Substrate (1705061) or Clarity Max Western ECL Substrate (1705062) from Biorad (CA, United States), and the images were captured with the Azure Biosystem Aerogene 300Q imaging system. Densitometric analysis was performed using ImageJ software: after background subtraction, the intensity of the signal of each band of the protein of interest was normalised to the signal of the corresponding loading control (GAPDH or VDAC1).

Primary antibodies included rabbit anti-PITRM1 (1:500, Atlas Antibodies #HPA006753, Sweden), rabbit anti-Frataxin (1:1,000, Abcam #ab175402, UK), rabbit anti-IDE (1:1,000, Abcam #ab228720,UK), rabbit anti-PMPCB (1:1,000, Abclonal #A4312, MA, United States), rabbit anti-GAPDH (1:1,000, Abcam #ab181602, UK) and rabbit anti-VDAC1 (1:1,000, Abcam #ab15895, UK); Rabbit Polyclonal PPARgamma/NR1C3 (1:100, Bio-techne #NB120-19481, MI, United States), Total OXPHOS Human WB antibody cocktail (1:500, Abcam, #ab110411), mouse anti-TFAM (1:1000, ThermoFisher MA5-16148). Secondary antibodies include ECL anti-mouse IgG peroxidase (Sigma-Aldrich, GENA931) and ECL anti-rabbit IgG peroxidase (Sigma-Aldrich, GENA9340).

To quantify mtDNA, we collected cell pellets and extracted total DNA, using QIAamp DNA Micro Kit from Qiagen (#56304). Real-time quantitative PCR (qPCR) was performed by using SsoAdvanced universal Probes Kit (BioRad, #1725281), with primers and probe for the 12S subunit of mitochondrial rRNA (MT-RNR1) and TaqMan RNAse P control reagent Kit (Applied Biosystem, #4316844) as nuclear reference (Nasca et al., 2022). We performed three independent qPCR analysis for each cell line and for each condition. Primers sequences are listed in Table 2.

To detect the accumulation of MTS through cell imaging we designed a dedicated fluorescent indicator based on biarsenical-binding tetracysteine motifs (TC). It has been shown that the (cell permeable) biarsenical reagents FlAsH-EDT2 can bind TC motifs. Upon binding, a thiol-arsenic ligand exchange reaction converts these nonfluorescent reagents into fluorescent, protein-bound, complexes (Griffin et al., 1998; Jakobs, 2006). We introduced a TC motif into the targeting sequence of the mitochondrial matrix protein TFAM, fused to the monomeric far-red fluorescent protein mKATE (Shcherbo et al., 2007; 2009). To design the probe, we first analysed the primary structure of TFAM by means of bioinformatic approaches to define the length of the MTS and predict the cleavage site of the mitochondrial matrix peptidase. According to MitoFates (Fukasawa et al., 2015) TPpred 2.0 (Savojardo et al., 2014) and TargetP 1.1 (Nielsen et al., 1997; Emanuelsson et al., 2000), the latter resides at amino acid 42, yielding the following MTS: MAFLRSMWGVLSALGRSGAELCTGCGSRLRSPFSFVYLPRWF. Implementation of the TC motif within this sequence was not random, as its insertion should not affect the properties of the subcellular localization sequence (e.g., efficiency of mitochondrial targeting) and of the cleavage site of the processing peptidase. We chose the smallest TC sequence (CCPGCC) sufficient for efficient FlasH binding as shown in (Griffin et al., 1998; Jakobs, 2006). We inserted it at position 21 after considering that the FlasH binding could take advantage not only of the TC motif that we inserted without changing the overall length of the MTS, but also of its flanking amino acids (Martin et al., 2005; Luedtke et al., 2007). We then confirmed though bioinformatic analyses that the new TFAM-MTS (which reads as follow: MAFLRSMWGVLSALGRSGAECCPGCCLRSPFSFVYLPRWF) was still able to drive localization within mitochondria (YLOC (Briesemeister et al., 2010a; Briesemeister et al., 2010b), iPSORT (Bannai et al., 2002), MitoFates, Target P 1.1, TPpred 2.0; scored respectively 68%, yes, 97%, 98%, 99% probabilities of mitochondrial localization/presence of peptidase cleavage site at position 42). A custom sequence including human TFAM (NCBI Reference Sequence: NP_003192.1), modified in its MTS as reported above and fused at its C-terminus with mKATE was inserted into the pUC57-Amp plasmid (gene synthesis from GENEWIZ, GENEWIZ South Plainfield, NJ). The fragment encoding for the TFAM-Kate was PCR amplified by using the forward (GGA​GGA​TCC​ATG​GCC​TTC​CTG​AG) and reverse (CCT​CTC​GAG​TCA​TCA​CCT​GTG​GC) primers, digested with BamHI and XhoI and cloned into the BamHI-XhoI sites of the pcDNA™5/FRT vector (Supplementary Figure S1). All restriction and modification enzymes were from Fermentas (St. Leon-Rot, Germany) unless otherwise stated.

To label TC-tagged TFAM-Kate overexpressed in fibroblasts, the TC-FlAsH II In-Cell Tetracysteine Tag Detection Kit from Molecular Probes (Thermo Fisher Scientific) was utilized. The transfected CTRL B and PT3 cells were cultured in 35 mm glass-bottom dishes. For monitoring the effects over time, in-cell FlAsH labeling was conducted 48 h post-transfection. The cells were washed twice with reduced serum Opti-MEM without phenol red (Life Technologies) and then incubated with FlAsH at a concentration of 1 μM in Opti-MEM. This incubation took place at 37°C for 30 min, while being protected from light. After 30 min, the cells were washed twice with 1X BAL (2,3-dimercaptopropanol) buffer in Opti-MEM, again at 37°C and protected from light, for 1 h. Cells were stained with Hoechst (1 μg/mL) in Opti-MEM for 5 min. Finally, the cells were washed with Opti-MEM and prepared for live-cell imaging. The samples were visualized using an inverted laser scanning microscope (Axio Observer.Z1). Fiji software was employed for image analysis. Flash fluorescence images were segmented by automated color thresholding, enabling the identification of fluorescent positive particles within the cell cytoplasm. The average grey value of each particle was measured using the “analyze particles” command. Finally, the mean particle intensity per cell was calculated and used for statistical analysis.

All results are expressed as mean ± standard deviation (SD). Graph Pad Prism, version 7.03 software was used to generate the graphs and to perform statistical analyses, which are indicated in the corresponding figure legend.

To evaluate the pathogenic effect of the PITRM1^R183Q, PITRM1^V747Gfs*31 and PITRM1^T931M variants (all together referred to as PITRM^MUT), we first modelled the 3D structures of wild-type (WT) and mutated versions of PITRM1 using Pymol software.

The in silico docking analysis of mutated protein versus Aβ peptide (a PITRM1 substrate) predicts that all three PITRM1 variants do not affect the catalytic site, suggesting that the PITRM1 enzymatic activity is preserved (Figure 1A). The PITRM1 transcript levels were similar to control in PITRM1^R183Q (PT1) and PITRM1^T931M (PT3), but slightly reduced in PITRM1^V747Gfs*31 (PT2) fibroblasts, although this difference was not statistically significant (fold changes: CTRLs: 100 ± 2.8; PT3: 46.3 ± 11, p > 0.05, Figure 1B). However, Western blot analysis of proteins separated by SDS–PAGE showed a marked reduction of PITRM1 amount in all patients (fold changes: CTRLs: 100 ± 23; PITRM1^MUT: 12.9 ± 6.2, p < 0.005, Figures 1C, D), suggesting that all the three variants affect protein stability.

Previous works done in yeast (Mossmann et al., 2014) and in human PITRM1 KO cell lines (Kücükköse et al., 2021) showed that PITRM1 deficiency leads to the build-up of non-degraded MTSs that accumulate within the mitochondrial matrix, causing dissipation of the ΔΨm and impairing the processing of presequence proteins by the peptidase MPP.

To verify whether this proposed pathogenic mechanism is present also in the mutant patients, we evaluated ΔΨm by JC1 staining in PT 1, PT 2, and PT 3 fibroblasts (Figure 2A). The mitochondrial membrane potential was significantly compromised in PITRM1^MUT cell lines, which displayed a higher proportion of depolarised organelles compared to controls (depolarised/healthy mitochondria: CTRLs: 13.7%/85.3% ± 6.03%; PITRM^MUT: 55.3%/44.7% ± 4.1%, p < 0.005, Figure 2B). This result confirms previous observations described in PITRM1^R183Q (Brunetti et al., 2016) and suggests that mitochondrial depolarisation is a common hallmark for PITRM1 pathogenic variants.

MTSs can bind to the mitochondrial membrane and disrupt the electrochemical gradient of the mitochondria (Ståhl et al., 2002); therefore, to assess whether MTSs scavenging is compromised when PITRM1 is dysfunctional, we transfected CTRL B and PITRM1^T931M mutant fibroblasts (PT3) with the TC-FlAsH-TFAM-Kate reporter vector (Figure 3A). This probe holds a TC motif into the MTS of the mitochondrial matrix protein TFAM, fused to the monomeric far-red fluorescent protein mKATE (Figure 3A). Degradation of MTS should impede the detection of the FlAsH signal, whereas its accumulation should result in enhanced FlAsH-TC fluorescence (Figure 3B).

To determine whether FlAsH labelling was able to identify MTS-TC, we compared live-cell images after the addition of the FlAsH reagent to MTS-TC-TFAM-Kate expressing cells 48 h post-transfection (Figure 3C).

We measured the mean fluorescence intensity of Flash-TC and found that the intensity of the FlAsH-TC (green) signal was higher in the PITRM1^T931M mutant cell line compared to CTRL B (fluorescence intensity (arbitrary unit, a.u.): CTRL B: 46.15 ± 9.02; PT3: 79.82 ± 12.79, p < 0.001, Figure 3D).

To assess whether MPP presequence processing is compromised when PITRM1 amount is reduced, we analyzed by SDS-PAGE the levels of the exogenous presequence containing TFAM-Kate precursor, using an anti-TFAM antibody (Figure 3E). While the TFAM-Kate was processed by MPP in the control cells leading to the detection of cleaved form mTFAM, the amount of mature protein was strongly reduced in PITRM1^T931M mutant cells (CTRL B: mTFAM/iTFAM: 0.32 ± 0.01; PT3: 0.18 ± 0.007, p < 0.01, Figure 3F).

Altogether, these findings indicate that the proposed pathomechanism described in PITRM1-deleted yeast or in PITRM1 KO cell models is also present in PITRM1^MUT fibroblasts.

Since mitochondrial peptidases are also involved in the maturation of the human Frataxin precursor, we examined Frataxin protein species, i.e., mature and immature forms (mFXN and iFXN, respectively), by immunoblotting in control versus patients fibroblasts (Figure 4A). PITRM1-mutant cells showed a significantly reduced amount of mature Frataxin (mFXN) compared to controls (fold changes: CTRLs: 100 ± 16.7; PITRM1^MUT: 54.39 ± 16.05, p < 0.05, Figure 4B), as well as a non-statistically significant increase of iFXN form (fold changes: CTRLs: 100 ± 29.7; PITRM1^MUT: 129 ± 33, p > 0.05, Figure 4C) which ultimately impaired the mFXN/iFXN ratio (CTRLs: 15.01 ± 4.29; PITRM1^MUT: 5.66 ± 2.24, p < 0.01, Figure 4D), indicating a decreased function of MPP and defects of mitochondrial presequence processing.

To understand whether targeting PITRM1 enzymatic activity by external stressors may reduce Frataxin processing and mitochondrial membrane potential, we treated the PITRM1^R183Q cells with H₂O₂, which is known to reduce PITRM1 activity (Chen et al., 2014). Furthermore, since IDE could compensate for the PITRM1 deficiency by degrading MTSs as reported in yeast (Mossmann et al., 2014), we induced a pharmacological inhibition of this activity by “engulfing” IDE’s enzymatic activity with Insulin supplementation, as IDE has a higher affinity to Insulin than to MTS.

As revealed by immunoblot, PITRM1 protein levels were significantly reduced in PITRM1^R183Q fibroblasts and were not further modulated by H₂O₂ or insulin treatments (Figure 5A, B); No obvious changes were detected in IDE protein levels in both the CTRL and PITRM1^MUT cells (Figures 5A, B). This notwithstanding, the amount of mFXN was significantly reduced by both treatments in PITRM1^R183Q but not in CTRL cells (Figure 5B). This suggests that exogenous stressors targeting either PITRM1 or IDE have more detrimental effects on PITRM1-deficient cell lines, resulting in a reduction of MPP activity. Moreover, the amount of PMPCB (an MPP subunit) did not change (Figures 5A, B), thus excluding that the observed processing defects were caused by reduced amount of protein, thus confirming previous results obtained by others on PITRM1 ^KO HEK293T cells (Kücükköse et al., 2021).

To better evaluate the impact of the impaired proteostasis exacerbated by the exogenous stressor agents on the mitochondria, we examined the mitochondrial network architecture (Figure 5C). First of all, untreated fibroblasts from PT 1 displayed mild yet significant alteration of the mitochondrial network, with a higher percentage of fragmented, and a lower percentage of elongated mitochondria compared to the control cell line (CTRL B: elongated/fragmented: 79.95%/20.05% ± 5.09%; PT1: elongated/fragmented 70.83%/29.17% ± 1.04%, p < 0.05, Figure 5D). The mitochondrial network architecture of control fibroblasts was not affected by insulin or H₂O₂ treatment (Figures 5C, D); on the contrary, PITRM1^R183Q cells displayed an increased network fragmentation with exogenous stressors (Figure 5C) which was exacerbated specifically by Insulin treatment (PT1 untreated: elongated/fragmented: 70.83%/29.17% ± 1.04%, insulin-treated PT1: elongated/fragmented: 66.50%/38.50% ± 3.27%, p < 0.01, Figure 5D).

Altogheter, these results suggest that oxidative stress and increased insulin levels are detrimental for PITRM1^MUT patients. The same deleterious effects may be expected during aging when the level of PITRM1 is lower (Brunetti et al., 2020).

Given that Ste23 (the homologue of human IDE in yeast), is required for efficient leader peptides degradation (Taskin et al., 2017), we assessed whether the pharmacological stimulation of IDE could act as a protective mechanism against MTS accumulation and defects of mitochondrial protein maturation. To this end, CTRLs and PITRM1^MUT fibroblasts were treated with Pioglitazone (PG), a PPARG agonist, with a reported effect on IDE upregulation. Indeed, Pioglitazone treatment (5 µM for 48 h) upregulated IDE levels improving Frataxin maturation specifically in PITRM1^MUT fibroblasts (fold changes PITRM1^MUT: vehicle-treated: 67.97 ± 52.72; PG-treated: 267.9 ± 94.7, p < 0.05, Figures 6A, B). Surprisingly, we also detected a strong upregulation of PITRM1 levels, both in CTRLs (fold changes: vehicle-treated: 100.32 ± 22.16; PG-treated: 298.98 ± 54.96, p = 0.058) and in mutated cells (fold changes: vehicle-treated: 22.49 ± 14.06; PG-treated: 273.78 ± 145.33, p < 0.001, Figure 7B). In parallel, the maturation of Frataxin has also been restored in patients’ fibroblasts upon Pioglitazone treatment (fold changes: vehicle-treated: 61.36 ± 8,866; PG-treated: 112.17 ± 27.32, p < 0.005). To evaluate whether the upregulation of IDE and PITRM1 provided by Pioglitazone treatment was associated to an improvement of the mitochondrial membrane potential, we measured ΔΨm using JC1 staining. As shown in Figure 6C, control fibroblasts showed a preponderant emission at 590 nm wavelength (red fluorescence) in both conditions, indicating the preservation of the mitochondrial membrane potential (depolarised/healthy mitochondria: vehicle-treated: 14.1%/85.83% ± 6.29%; PG-treated: 13.88%/86.11% ± 5.87%, p > 0.05, Figure 6D). On the contrary, patients fibroblasts presented with a diffuse green fluorescence confirming the presence of a defective ΔΨm (Figures 6C, D), which interestingly, it was corrected by Pioglitazone treatment (depolarised/healthy mitochondria: vehicle-treated: 55.44%/44.55% ± 4.47%; PG-treated: 24.97%/75.02% ± 12.01%, p < 0.005, Figure 6D) and became indistinguishable from control cells (Figures 6C, D).

Thus, we examined the expression levels of PPARG protein in basal conditions (only vehicle, VEH) and after PG supplementation (Figure 7A); we found that PPARG was significantly upregulated in patients’ fibroblasts after PG treatment (fold changes: vehicle-treated: 128.3 ± 64.24; PG-treated: 279.6 ± 231.2 p < 0.05, Figure 7B).

Since Pioglitazone has been proposed to activate mitochondrial biogenesis (Corona and Duchen, 2016) we evaluated whether the increase in PITRM1 protein level observed after pioglitazone treatment was related to an increase in mitochondrial content. However, no significant differences in mitochondrial DNA copy number (Figure 7C) or in mitochondrial OXPHOS proteins were observed in VEH- versus PG-treated cells from both controls and patients (Figures 7D, E). Only SDHB (subunit of complex II) was significantly reduced in patients’ fibroblasts upon PG treatment (fold changes: vehicle-treated: 148.4 ± 21.93; PG-treated: 57.91 ± 22.34, p < 0.05, Figure 7E).

Overall, our results suggest that PITRM1 gene is modulated by PPARG via a transcriptional mechanism. This hypothesis is supported by an in silico prediction revealing a significative binding score (p < 0.001) between PPARG and a nucleotide sequence (predicted as a PITRM1 enhancer) located at −813bp upstream the PITRM1 Transcription Starting Site (Figure 7F).