All procedures complied with the ethical guidelines of Sant Joan de Déu Children’s Hospital and were approved by the Clinical Research Ethics Committee under reference PIC-223-19. Informed consent was obtained from the patients, the patient’s parents, or legal guardians.

The pathogenicity classification of genetic variants was done using VarSome¹ (last accessed March 24, 2021) and the American College of Medical Genetics and Genomics (ACMG) standard guidelines. Genome Aggregation Database (gnomAD; last accessed March 24, 2021)² and CIBERER-Spanish Variant Server (CSVS; last accessed March 24, 2021)³ were used to evaluate the frequencies of the variants in global and Spanish populations, respectively. In silico analyses of the genetic variants were performed using: PROVEAN (Choi and Chan, 2015), FATHMM (Shihab et al., 2013), DANN (Quang et al., 2015), MutationTaster (Schwarz et al., 2014), and CADD (Rentzsch et al., 2019).

Skin biopsy was taken from the patient following standard clinical procedures (Zuber, 2002; Vangipuram et al., 2013). Healthy control fibroblasts were obtained from Sant Joan de Déu Children’s Hospital Biobank. Fibroblasts were cultured in Dulbecco’s Modified Eagles’ Medium high-glucose (Sigma-Aldrich, St. Louis, MO, United States) supplemented with 10% fetal bovine serum (FBS; Sigma-Aldrich), 2 mM L-glutamine (Sigma-Aldrich) and 100 mg/ml penicillin-streptomycin (Sigma-Aldrich) at 37°C in a 5% CO₂ incubator. Cells were periodically tested for mycoplasma infection.

Fibroblasts (3 × 10⁴) were seeded onto glass coverslips for 24 h, washed with phosphate buffer saline (PBS) and fixed in pre-warmed 4% paraformaldehyde (PFA) for 20 mins at room temperature. Fibroblasts were permeabilized with 0.2% Triton X-100 in PBS for 30 mins at room temperature and they were blocked with 1% bovine serum albumin (BSA) and 4% serum in PBS for 1 h at room temperature. The specific primary antibodies α-TOM20 mouse monoclonal (1:100; BD Transduction Laboratories, Franklin Lakes, NJ, United States; 612278), α-LAMP-1 mouse antibody (1:200; DSHB, Iowa, IA, United States; H4A3), α-LAMP-1 rabbit polyclonal (1:500; Abcam; ab24170) and α-p62/SQSTM1 rabbit polyclonal (1:200; Sigma-Aldrich; P0067) were incubated overnight at 4°C. Primary antibodies were visualized using Alexa Fluor^® 488 or Alexa Fluor^® 633-labeled secondary antibody (1:500; Thermo Fisher Scientific; A11034, A11029, and A21070). After 2 h of incubation, fibroblasts were rinsed with PBS and mounted on a coverslip using Fluoromont-G with DAPI (4′,6-diamidino-2-phenylindole) (Thermo Fisher Scientific).

Fibroblasts (3 × 10⁴) were seeded onto glass coverslips for 24 h, washed in PBS, fixed in pre-warmed 4% PFA for 20 mins at room temperature and permeabilized with ice-cold methanol at −20°C for 20 mins. After 1 h of incubation at 37°C with the blocking solution in a pre-heated humidity chamber, fibroblasts were incubated overnight at 4°C with the specific primary antibodies: α-GDAP1 mouse monoclonal (1:200, Abcam, Cambridge, United Kingdom; ab194493), α-MFN2 mouse monoclonal (1:100, Abcam; ab56889) and α-LAMP-1 rabbit polyclonal (1:100, Abcam; ab24170). Afterward, we perform the PLA assay according to the manufacturer’s instructions (Duolink^® In Situ Detection Red Starter [Mouse/Rabbit] Kit; Sigma-Aldrich) and the coverslips were mounted with Duolink In situ Mounting Medium with DAPI. Images were acquired with a Leica TCS SP8 X White Light Laser confocal microscope using 63× oil immersion objective and Z-stacks were acquired every 0.1 μm along with the cell thickness. For PLA with co-stained mitochondria and lysosomes, images were acquired using 100× oil immersion objective and deconvolution was performed with Huygens Essential software v 4.4 0p6 (SVI, Leiden, Netherlands). Image analysis was performed using maximum intensity projection in Image J/Fiji software (NIH, US National Institutes of Health, Bethesda, MD, United States). For each antibody, a negative control experiment was performed, where only one antibody was incubated with the PLA probes.

We analyzed the mitochondrial membrane potential and the mitochondrial oxidative stress with TMRM and MitoSOX probes, respectively. We measured these parameters using two different approaches: live-cell imaging and flow cytometry.

Fibroblasts were homogenized in lysis buffer (50 mM Tris HCl pH 7.4, 1.5 mM MgCl₂, 5 mM EDTA, 1% Triton X-100, 50 mM NaF, and 1 mM Na₂VO₃) containing a protease inhibitor cocktail (Complete Mini-Protease Inhibitor Cocktail, Roche, Branchburg, NJ, United States). Homogenates were centrifuged at 13,200 rpm for 15 min at 4°C and the protein concentration of the supernatant was quantified by the BCA method (Thermo Fisher Scientific). For immunoprecipitation assays, 1 mg of total lysate was incubated with the specific antibody for 6–8 h at 4°C followed by incubation with Protein G Sepharose™ 4 Fast Flow (GE, Healthcare) overnight at 4°C. Beads were softly washed with lysis buffer, resuspended in Laemmli Buffer, and heated at 95°C. Lysates and Co-immunoprecipitation (co-IPs) were resolved in sodium dodecyl sulfate-polyacrylamide gels (SDS-Page) and transferred onto Amersham Hybond PVDF membranes (GE Healthcare, Chicago, IL, United States). The membranes were blocked with 5% defatted-milk in TBS-0.1% Tween 20 buffer (25 mM Tris, 50 mM NaCl, 2.5 mM KCl, and 0.1% Tween-20). Afterward, the membranes were blotted with the specific primary antibodies α-MFN2 rabbit polyclonal (1:500; Sigma-Aldrich; M6319), α-MFN2 mouse monoclonal (1:1,000; Abcam; ab56889), α-GDAP1 rabbit polyclonal (1:500; HPA024334; Sigma-Aldrich), α-LAMP-1 rabbit polyclonal (1:500; Sigma-Aldrich; L1418), α-LAMP-1 mouse monoclonal (1:500; DSHB; H4A3), α-p62/SQSTM1 rabbit polyclonal (1:500; Sigma-Aldrich; P0067), α-LC3B rabbit polyclonal (1:500; Sigma-Aldrich, L7543), and α-β Actin mouse monoclonal (1:8,000 5% BSA; Sigma-Aldrich; A5316) which were detected using secondary antibodies coupled to horseradish peroxidase. Proteins were processed for chemiluminescence with Amersham ECL Prime Western Blotting Detection Reagent and visualized by iBright™ CL1000 Imaging System (Thermo Fisher Scientific).

Super-resolution images were acquired with a Leica TCS SP8 X White Light Laser confocal microscope with Hybrid spectral detectors and HyVolution (Leica Microsystems, Wetzlar, Germany) using the Leica LAS X software (version 3.1.5). Images were acquired using 100× or 63× oil immersion objectives. The original data was stored as 16-bit greyscale images with a spatial resolution of 1,024 × 1,024 pixels. Z-stacks were acquired every 0.16 μm along with the cell thickness. Image processing and analysis were performed using Image J/Fiji software (NIH, US National Institutes of Health) and the Leica Application Suite X (LAS-X) software (Leica Microsystems). To compare the data, identical settings were used for image acquisition of different experiments and negative control samples were used for background setting previous to image acquisition.

All data are expressed as mean ± standard deviation (SD) or box plots showing the median, box edges represent the [25th and 75th percentiles], and the whiskers extend to the minimum and maximum values. The normality of data was assessed by the Kolmogorov-Smirnov test. Statistical analysis was performed using GraphPad Prism (version 8.0.1; GraphPad Software, Inc., La Jolla, CA, United States) with a minimum of three independent experiments. The specific test applied in each case is indicated in the figure legend. P-values less than 0.05 were considered significant. P-values are indicated by asterisks *P < 0.05, ^**P < 0.01, ^***P < 0.001.

To discover cellular markers related to the effect of pathogenic variants in neurogenetic diseases, we included seven patients affected by mutations of different Mendelian disorders that are associated with mitochondrial phenotypes. We ascertained the patients based on two categories: (i) four involving genes related to mitochondrial network dynamics, either the fission process, DRP1 (also DNM1L, encoding dynamin-related protein 1) and GDAP1 (ganglioside induced differentiation-associated protein 1) or the fusion process, OPA1 (optic atrophy protein 1) and MFN2 (mitofusin 2), and (ii) three whose the gene dysfunction affects mitochondria or induces a mitochondrial phenotype but are not directly involved in mitochondrial dynamics, FXN (frataxin, mitochondrial matrix) (Lynch and Farmer, 2021), MED13 (mediator complex subunit 13, a transcriptional coactivator that prevents mitochondrial fission and programmed cell death) (Cooper et al., 2014; Khakhina et al., 2014) and CHKB (choline kinase beta, phospholipids synthesis) (Mitsuhashi et al., 2011). We provided the clinical and genetic features (Table 1), the in silico analysis of the genetics variants (Table 2) and their location in the encoded proteins (Figure 1).

First, we studied in patients’ fibroblasts the mitochondrial morphology by immunostaining TOMM20 (OMM marker) and quantified mitochondrial mass, elongation and fragmentation. We observed in DRP1^K75E/+ fibroblasts a peculiar network with a “pearl-chain-like” structure, which is a cellular phenotype of patients with DRP1 variants (Nasca et al., 2016), GDAP1^W67L/W67L had a tangled network, OPA1^F570L/+ and MFN2^R104W/+ network were fragmented, FXN^R165C/GAA showed a thick pattern, and MED13^L830R/+ showed an elongated network. In contrast, CHKB^Q198*/Q198* had a similar network when compared to control fibroblasts (Figure 2A).

Regarding quantitative parameters, the mitochondrial mass showed a significant decrease in GDAP1^W67L/W67L, OPA1^F570L/+, and MED13^L830R/+ fibroblasts, and a significant increase in FXN^R165C/GAA (Figure 2B). The inverse average circularity showed that mitochondria were highly elongated in GDAP1^W67L/W67L, OPA1^F570L/+, and MED13^L830R/+ fibroblasts while the opposite was found in FXN^R165C/GAA (Figure 2C). The fragmentation of the mitochondrial network was increased in OPA1^F570L/+ and MFN2^R104W/+ (proteins involved in mitochondrial fusion) while a significant decrement was found in FXN^R165C/GAA and MED13^L830R/+ fibroblasts (Figure 2D). Taking into account all these results, we categorized the mitochondrial network shape of each patient (Figure 2D). We determined that mitochondrial networks of patients’ fibroblasts had different shapes (Figure 2E). The network was entangled in GDAP1^W67L/W67L, fragmented in both OPA1^F570L/+ and MFN2^R104W/+, elongated in MED13^L830R/+, thick in FXN^R165C/GAA and similar to control fibroblasts in CHKB^Q198*/Q198*. These analyses in fibroblasts show changes in mitochondrial network morphology that are gene-specific except for CHKB^Q198*/Q198* fibroblasts.

Since mitochondrial network dysfunction is associated with defects in mitochondria bioenergetics, and the generation of ROS (Benard et al., 2006; Kausar et al., 2018), we wondered if changes in the mitochondrial network of patients’ fibroblasts could also affect the mitochondrial membrane potential (ΔΨm) and generate ROS. The mitochondrial membrane potential was analyzed by live-cell imaging and flow cytometry using TMRM probe (Figures 3A–C). The comparison with control fibroblasts revealed significant differences only in MED13^L830R/+ fibroblasts, which showed an increase in ΔΨm by flow cytometry but not in live-cell imaging. In contrast, we found mitochondrial oxidative stress in all the patients using MitoSOX probe with both technical approaches (Figures 3D–F). These results show that oxidative stress is a common cellular phenotype and a pathological marker in all patients without a decrease in ΔΨm.

In a previous study, we have reported that GDAP1 participates in mitochondria–lysosome contacts by interacting with LAMP-1 (lysosome-associated membrane protein-1, a marker of lysosomes) (Cantarero et al., 2020). Here, we investigated the role of these contacts in all patients’ fibroblasts using proximity ligation assays (PLA). We found a significant decrease in the number of PLA dots in both GDAP1^W67L/W67L and MFN2^R104W/+ fibroblasts (Figures 4A,B). Since GDAP1 and MFN2 are located in the OMM, we hypothesized that such a reduction could be caused by the interaction of these proteins. Co-immunoprecipitation assays revealed a constitutive interaction between GDAP1 and MFN2 (Figure 4C). To further investigate the relationship between these proteins, we asked if MFN2 interacts with LAMP-1 in mitochondria–lysosome contacts. Co-IP and PLA experiments revealed the constitutive interaction between MFN2 and LAMP-1, localizing PLA dots where both organelles are present (Figures 4D,E). Moreover, we found in both GDAP1^W67L/W67L and MFN2^R104W/+ fibroblasts a significant reduction in the MFN2–LAMP-1 interaction (Figure 4F). Importantly, this reduction occurred without a decrease in GDAP1, MFN2, or LAMP1 expression (Figure 4G), supporting that missense variants in GDAP1^W67L/W67L and MFN2^R104W/+ affect mitochondria–lysosome contacts. These results argue that MFN2 participates in mitochondria–lysosome MCSs and that the decrease in these contacts is a unique feature of patients’ fibroblasts carrying pathogenic variants in OMM genes.

Given that mitochondrial biology and mitochondria-lysosome MCSs have been reported to regulate lysosomal morphology and dynamics (Demers-Lamarche et al., 2016; Wong et al., 2018; Cantarero et al., 2020), we immunostained patients’ fibroblasts using α-LAMP-1 (Figures 5A,B). First, we analyzed GDAP1^W67L/W67L and MFN2^R104W/+ fibroblasts that showed a reduction in mitochondria-lysosome MCSs (Figure 5). We found a significant increment in the lysosomal area in GDAP1^W67L/W67L but not in MFN2^R104W/+ that was similar to control fibroblasts. The lysosomal area was also increased in DRP1^K75E/+, OPA1^F570L/+, and FXN^R165C/GAA fibroblasts. Furthermore, we observed a correlation between the lysosomal area increment and its distribution that was abnormal in DRP1^K75E/+, GDAP1^W67L/W67L, OPA1^F570L/+, and FXN^R165C/GAA fibroblasts (Figure 5A). These results show a relationship between defects in lysosomal morphology and their distribution in fibroblasts and suggest lysosomal dysfunction.

To continue examining the mitochondria-lysosome axis, we studied the autophagic flux in patients’ fibroblast with mutations in genes of mitochondrial dynamics (DRP1^K75E/+, GDAP1^W67L/W67L, OPA1^F570L/+, and MFN2^R104W/+). We performed a western blotting analysis of sequestosome-1/p62 (autophagic flux) and LC3-II/LC3-I ratio (a marker for autophagosome formation) (Bjorkoy et al., 2009) in untreated cells and after Bafilomycin A1 (BafA1) treatment (inhibits autophagosome-lysosome fusion) (Figure 6A). We found an increase of both sequestosome-1/p62 and LC3-II/LC3-I ratio in all samples except for GDAP1^W67L/W67L, which showed a non-significant increase. This result in GDAP1^W67L/W67L is consistent with previous findings in GDAP1-deficient models (Cantarero et al., 2020). Subsequently, we analyzed in all patients the aggregates forming of sequestosome-1/p62 by immunofluorescence in cells treated with BafA1 or Earle’s balanced salt solution (EBSS), which causes starvation and promotes autophagy. In basal conditions (Figure 6B, upper panel), α-p62 immunostaining experiments revealed qualitative differences among patients regarding the size and number of p62 aggregates. Besides, FXN^R165C/GAA and MED13^L830R/+ showed a significant accumulation of these aggregates (Figure 6C). After the treatments control fibroblasts showed the physiological behavior of the expected autophagic flux: BafA1 significantly increased the number of p62 aggregates per cell, while starvation had the opposite consequence with a reduction of the number of p62 aggregates (Figures 6B, medium and lower panels, and 6D). In patients’ fibroblasts, quantification of p62 aggregates resulted in four patterns of response to treatments (Figure 6D): (i) total absence of response in DRP1^K75E/+, (ii) positive response to BafA1 with no response to EBSS in GDAP1^W67L/W67L, OPA1^F570L/+, and MFN2^R104W/+, but also CHKB^Q198*/Q198*, (iii) no response to BafA1 with a positive response to EBSS in MED13^L830R/+ and (iv) positive response to treatments in FXN^R165C/GAA, similar to the response of control cells. These results show variability among patients in terms of autophagic flux impairment and/or autophagy induction by starvation.