ISCA1 mutation in a patient with infantile-onset leukodystrophy causes defects in mitochondrial [4Fe–4S] proteins

Abstract

Multiple mitochondrial dysfunction syndromes (MMDS) comprise a group of severe autosomal recessive diseases characterized by impaired respiration and lipoic acid metabolism, resulting in infantile-onset mitochondrial encephalopathy, non-ketotic hyperglycinemia, myopathy, lactic acidosis and early death. Four different MMDS have been analyzed in detail according to the genes involved in the disease, MMDS1 (NFU1), MMDS2 (BOLA3), MMDS3 (IBA57) and MMDS4 (ISCA2). MMDS5 has recently been described in a clinical case report of patients carrying a mutation in ISCA1, but with no further functional analysis. ISCA1 encodes a mitochondrial protein essential for the assembly of [4Fe–4S] clusters in key metabolic and respiratory enzymes. Here, we describe a patient with a severe early onset leukodystrophy, multiple defects of respiratory complexes and a severe impairment of lipoic acid synthesis. A homozygous missense mutation in ISCA1 (c.29T>G; p.V10G) identified by targeted MitoExome sequencing resulted in dramatic reduction of ISCA1 protein level. The mutation located in the uncleaved presequence severely affected both mitochondrial import and stability of ISCA1. Down-regulation of ISCA1 in HeLa cells by RNAi impaired the biogenesis of mitochondrial [4Fe–4S] proteins, yet could be complemented by expression of wild-type ISCA1. In contrast, the ISCA1 p.V10G mutant protein only partially complemented the defects, closely resembling the biochemical phenotypes observed for ISCA1 patient fibroblasts. Collectively, our comprehensive clinical and biochemical investigations show that the ISCA1 p.V10G mutation functionally impaired mitochondrial [4Fe–4S] protein assembly and hence was causative for the observed clinical defects.

Introduction

The biogenesis of mitochondrial iron–sulfur (Fe–S) proteins is an intricate, stepwise process. It involves 18 known proteins of the iron–sulfur cluster assembly (ISC) machinery (1), a highly conserved system amongst species from bacteria to man (2). Five major steps of the assembly process can be distinguished: (i) synthesis of a [2Fe–2S] cluster on the scaffold protein ISCU, (ii) cluster transfer to the monothiol glutaredoxin-5 (GLRX5), (iii) insertion into [2Fe–2S] recipient proteins; (iv) conversion of the [2Fe–2S] to a [4Fe–4S] cluster and (v) cluster transfer and insertion into [4Fe–4S] target apoproteins. Inactivation of ISC factors causes embryonic lethality in mammals (3–5). Particularly, the last two steps of the ISC pathway are still poorly characterized. Key players in the formation of [4Fe–4S] clusters are ISCA1, ISCA2 and IBA57 that physically interact with each other, and facilitate the conversion of the GLRX5-bound [2Fe–2S] into a [4Fe–4S] cluster in a biochemically ill-defined reaction (6–10). Three additional proteins, namely NFU1, IND1 (NUBPL) and BOLA3, are presumably involved in the transfer and insertion of the [4Fe–4S] clusters into mitochondrial apoproteins such as the respiratory chain complexes (RCCs) I and II, lipoate synthase (LIAS) and aconitase (ACO2) (11,12).

Mutations in many ISC genes are associated with severe mitochondrial diseases displaying a wide range of phenotypes spanning from anemia, mitochondrial encephalopathy, myopathy, and non-ketotic hyperglycinemia (13–15). In particular, mutations in the late-acting ISC genes are responsible for the Multiple Mitochondrial Dysfunction Syndromes (MMDS), characterized by white matter abnormalities, failure to thrive, feeding difficulties, neurologic regression, epilepsy and multiple respiratory chain defects (16). Five different MMDS are known: MMDS1 (NFU1, OMIM #605711) (17,18), MMDS2 (BOLA3, #614299) (18,19), MMDS3 (IBA57, #615330) (20,21), MMDS4 (ISCA2, #616370) (15) and MMDS5 (ISCA1, #617613) (22). For the latter gene, only a case report has been described but a detailed biochemical investigation of this disorder has not been performed leaving open the molecular explanation of the disease phenotypes. Herein, we report on a new case of MMDS5 and provide a comprehensive investigation of the clinical and biochemical consequences. The mutation in the ISCA1 gene results in early-onset leukodystrophy, severe spastic ataxia, lactic acidosis and respiratory defects. Biochemically, the p.V10G mutation located in the uncleaved mitochondrial presequence affects import, stability and function of ISCA1. These combined effects result in a general assembly defect of mitochondrial [4Fe–4S] proteins and explain the clinical features of the MMDS5 patients.

Results

Clinical features and brain nuclear magnetic resonance findings

The patient was the third boy of healthy non-consanguineous parents. He was born by cesarean delivery at the 40th week of gestation, with a birth weight of 3740 g. He was admitted to our hospital for acute urinary retention at the age of 6 years. In the family history, the first brother at the age of 6 months showed signs of a visual defect and early-onset developmental motor and cognitive delay; he died at the age of 1 year following an intercurrent infectious event. The eldest sister was healthy. The patient, similarly to his first brother, showed early loss of head control at the age of 3 months, nystagmus, and weak sucking with impairment in finishing his meals. At the age of 11 months, following a first MRI (data not shown), he was diagnosed to be affected by leukodystrophy and this pattern was confirmed later at the age of 4 years by a serial brain MRI (Fig. 1A–H). The MRI showed a pattern of vacuolating leukodystrophy. At the age of 18 months he was able to attain head control again, when sitting with support, and later at the age of 4 years he achieved the ability to sit without support. The neurological examination at the age of 6 years showed a child that was able to speak in sentences, with dysarthria, and was able to sit with some support showing a severe spastic ataxic syndrome, with persistent nystagmus. Brainstem-evoked potentials disclosed the absence of the V wave bilaterally and the delay of the I–III interpeak. The urodynamic examination showed a pattern of detrusor sphincter dyssynergia explaining the severe urinary overflow incontinence owing to a neurogenic bladder. One year later for worsening of swallowing difficulties the child was treated with a Nissen fundoplication gastrostomy. Pulmonary hypertension was ruled out by echocardiography. At the age of 8 years laboratory examinations showed metabolic acidosis with a base excess (BE) in blood of −9.0 mEq (n.v.: −2 to +2 mEq), low serum bicarbonate (14.6 mM; n.v.: 19–27 mM), no glycosuria, and normal values of creatine and blood urea nitrogen. Urinary organic acids as well as urinary amino acids were normal, ruling out increased glycine in urine. Lactate was only slightly increased in plasma to 2.8 mM (n.v.: 0.9–1.8 mM). The boy was treated with bicarbonate supplementation for several months, and died at the age of 11 years during an intercurrent pneumonia.

Figure 1.

Brain MRI and genetic features of the ISCA1 patient. Brain MRI performed at the age of 4 years shows a pattern of vacuolating leukodystrophy with abnormal T2 hyperintense areas that involve the white matter of the brain sparing the U fibers and the cerebellum (C, E, G, H); these abnormal areas are T1 hypointense (D) indicating vacuolization of the white matter. Of note, corpus callosum is thin and involved, as shown by T1 hypointense cavitatation areas (B) and hyperintense abnormalities are also evident in the pons (F, white arrow). Sagittal images weighted at T2 (A, C) and T1 (B, D); axial images weighted at T2 (E–G) and FLAIR (H). (I) Electropherograms of the genomic region of the patient (Pt) and the parents showing the c.29T>G variant in exon 1 of ISCA1. (J) Protein sequence alignment (ClustalW) highlights the mutation (p.V10G; red) in the putative presequence of human ISCA1. V10 is conserved only in vertebrates. The N-terminal pieces for Caenorhabditis elegans through Schizosaccharomyces pombe have been omitted.

RCCs activities of our patient were measured on autoptic muscle and brain specimens (autopsy was performed 1 h after death) that revealed multiple defects in both tissues (Supplementary Material, Table S1). Indirect evaluation of the OXPHOS status in mitochondria from patient fibroblasts by measuring the ATP synthesis, showed a 67% reduction with succinate, 31% with malate and 26% with pyruvate plus malate as substrates, pointing toward multiple defects of the RCCs (Supplementary Material, Table S2).

MitoExome sequencing revealed a new homozygous mutation in the ISCA1 gene

Bioinformatic analysis carried out on a MitoExome targeted panel identified five candidate genes for our patient, i.e. ATAD3B (NM_031921.5), ISCA1 (NM_030940.3), MMACHC (NM_015506.2), AOX1 (NM_001159.3) and TIMM44 (NM_006351.3). ATAD3B, MMACHC, AOX1 and TIMM44 were excluded to be involved in the disease pathogenesis by segregation analysis and because the variants did not follow a recessive trait of inheritance, hence further analysis of ISCA1 was pursued. Sanger sequencing confirmed the presence of the c.29T>G (p.V10G) variant in a homozygous state in the patient and in a heterozygous state in both healthy parents (Fig. 1I). The mutation was located in Exon 1 and was not reported in public databases [dbSNP (http://www.ncbi.nlm.nih.gov/sites/), ExAC (http://exac.broadinstitute.org/), EVS (http://evs.gs.washington.edu/EVS/)] and in in-house databases. In silico analysis with Polyphen-2 (PPH2, http://genetics.bwh.harvard.edu/pph2/), SIFT (http://sift.jcvi.org) and mutation taster (http://www.mutationtaster.org) predicted a harmful effect of the p.V10G variant which affects a residue moderately conserved (Fig. 1J). Moreover, mitochondrial localization prediction tools, e.g. MitoFates (http://mitf.cbrc.jp/MitoFates/cgi-bin/top.cgi) and TPpred2.0 (http://tppred2.biocomp.unibo.it/tppred2) suggested that the p.V10G mutation is part of the pre-sequence of the protein, potentially affecting the import of ISCA1 into mitochondria.

The ISCA1 p.V10G mutation is associated with decreased ISCA1 protein levels and multiple RCC defects

To evaluate the impact of the p.V10G mutation, we performed western blot (WB) analysis on mitochondria isolated from fibroblasts and found a dramatic reduction of ISCA1 levels in the patient compared with control cells (Fig. 2A). ISCA2, its interaction partner, was diminished although to a lesser extent, whereas IBA57 was not affected, in accordance with the findings showing that ISCA1 and ISCA2 form a tight hetero-complex (9).

Figure 2.

Decreased levels of ISCA1 and [4Fe–4S] cluster-containing proteins in ISCA1 patient fibroblasts. (A, B) Mitochondria isolated from control (C) and ISCA1 patient (Pt) fibroblasts were separated on a 12% SDS-PAGE and proteins were visualized by western blot (WB) analysis using specific antibodies against IBA57, ISCA1, ISCA2 and VDAC. The same samples were tested for the expression of subunits of the RCCI (NDUFS1), RCCII (SDHB), RCCIII (UQCRC2), RCCIV (COXIV), RCCV (F1β) and VDAC. The levels of the different proteins were normalized to VDAC levels (corresponding histograms). Data are presented as a mean ± SD of at least three independent experiments. (C) Fibroblast mitochondria were lysed with a mild detergent (lauryl maltoside), run on a BNGE and processed for WB analysis, using specific antibodies (RCCI-NDUFA9; RCCII-SDHB; RCCIII-UQCRC2; RCCIV-COXIV; RCCV-F1β). The same samples were loaded on 12% SDS-PAGE and probed with VDAC as control for equal loading. (D) Total cell lysates from controls and patient were processed for WB analysis to evaluate the amount of the mitochondrial (ACO2) and cytosolic (ACO1) aconitase. ACO2 expression was normalized against VDAC and TOMM20, whereas ACO1 was normalized against β-ACT and GAPDH (histogram). Data are presented as a mean ± SD of ≥3 independent experiments; *P< 0.05 and **P < 0.01.

Evaluation of the expression levels of different subunits of the RCCs showed a specific reduction of [4Fe–4S] cluster-containing RCCI (NDUFS1) and RCCII (SDHB), while the levels of other RCCs subunits were hardly changed (Fig. 2B). Blue Native Gel Electrophoresis (BNGE) followed by WB analysis confirmed the severe RCCI and RCCII diminution, as well as the unchanged levels of RCCIV and RCCV (Fig. 2C). We also noted a decrease of the RCCI + RCCIII super-complex level. This effect was fully caused by RCCI diminution, because the level of isolated RCCIII was even increased in the patient compared with VDAC (porin) levels (Fig. 2B and C). Collectively, these results were compatible with an impaired assembly of the [4Fe–4S] cluster-containing RCCI and RCCII, yet no major effects on the [2Fe–2S] cluster-containing RCCIII. The data are therefore in line with the dedicated function of ISCA1 (and ISCA2) in the assembly of mitochondrial [4Fe–4S] proteins (6,8).

The ISCA1 p.V10G mutation affects [4Fe–4S] but not [2Fe–2S] cluster-dependent enzymes

Based on the critical function of both ISCA1 and ISCA2 in the maturation of mitochondrial [4Fe–4S] proteins, we determined the protein level of mitochondrial aconitase (ACO2) as an indirect measure of the enzyme’s maturation status and found a slight decrease in ISCA1 patient cells (Fig. 2D, lower panel). Previous studies have revealed that an impaired mitochondrial function caused by ISCA1 or ISCA2 deficiency might also indirectly feed back to the proper maturation of cytosolic aconitase (ACO1/IRP1) (6). Consistently, we observed a 2- to 3-fold drop in ACO1 steady-state protein levels in patient fibroblasts (Fig. 2D, upper panel).

A critical enzyme related to mitochondrial energy metabolism is the [4Fe–4S] radical SAM enzyme lipoic acid synthase (LIAS) that catalyzes the final step in the formation of the lipoyl cofactor before its attachment to the E2 subunits of the pyruvate dehydrogenase complex (PDHc), the α-ketoglutarate dehydrogenase complex (KGDHc) and the α-branched-chain keto acid dehydrogenase complex (BCKDHc). Immunoblotting revealed a decrease in LIAS protein levels within ISCA1 patient mitochondria relative to controls, consistent with a destabilization of the apo-form of LIAS in patient cells (Fig. 3A). When we analyzed the lipoylation status of PDH-E2 and KGDH-E2 subunits by immunostaining of protein-bound lipoic acid (6), we found a severe drop in patient cells (Fig. 3B). This was accompanied by decreased levels of the PDHc subunits PDH-E2 (DLAT) and E3bp but not of PDH-E1α/β (Fig. 3C). Subsequently, we measured the enzyme activities of PDHc, KGDHc and BCKDHc in fibroblast mitochondria of patient and controls. In line with the impaired lipoylation of E2 subunits, we found reduced activities of all three enzymes in the patient sample (Fig. 3D), highlighting the critical consequences of the ISCA1 p.V10G mutation for the patient’s protein lipoylation and hence energy and metabolic homeostasis.

Figure 3.

Effect of the ISCA1 mutation on the protein lipoylation status. (A) LIAS level was assayed by WB analysis of mitochondria isolated from control (C) and ISCA1 patient (Pt) fibroblasts (left), and immuno-reactive bands were normalized to VDAC (right). (B) The lipoylation status of PDHc and KGDHc was determined by WB using a specific antibody against the lipoic acid of the E2 subunits (LA-E2; left). TOMM20 was used as a loading control and for quantitation (right). (C) The reduced lipoylation was associated with a decreased expression of the E2 (DLAT) and E3bp proteins of PDHc when compared with the expression of OSCP, an RCCV subunit. Data reported in the histograms of A−C are presented as a mean ± SD of ≥3 independent experiments. (D) Effects of the ISCA1 mutation on the enzyme activities of three mitochondrial lipoylated dehydrogenases (PDHc, KGDHc, BCKDHc). The assays were performed on samples from above by following the NADH formation at 340 nm. The data represent the mean ± SD of three independent experiments in duplicate; *P< 0.05; **P < 0.01 ***P < 0.001.

The ISCA1^V10G mutant protein cannot fully complement ISCA1-deficient HeLa cells

In order to analyze the consequences of the ISCA1 p.V10G mutation, we made use of a well-established HeLa cell culture model and analyzed the ability of the ISCA1^V10G mutant protein to complement the defects resulting from RNAi-mediated ISCA1 depletion (6). After 6 days of siRNA treatment (siISCA1) the endogenous ISCA1 protein levels were decreased by more than 80%, whereas the other late-acting ISC components IBA57, NFU1 and IND1 like mitochondrial MIA40 remained unaffected (Fig. 4A and B, Supplementary Material, Fig. S1A and B). ISCA1 protein levels were completely recovered by co-transfection of a plasmid encoding an RNAi-resistant version (smISCA1^wt). In contrast, expression of ISCA1^V10G from a plasmid containing the patient mutation (smISCA1^V10G) was less efficient and raised cellular ISCA1 levels only to one third of control cells (Fig. 4A and B). Consistently, smISCA1^wt reverted the siISCA1-associated growth defect, but expression of smISCA1^V10G did not (Supplementary Material, Fig. S2A). The phenol red indicator of the culture medium revealed a strong acidification upon ISCA1 depletion (Supplementary Material, Fig. S2B), indicative of a severe respiratory defect (6). Medium acidification was fully prevented upon expression of smISCA1^wt, but only incompletely counteracted in cells expressing variant smISCA1^V10G.

Figure 4.

The p.V10G mutation decreases ISCA1 protein levels and cannot fully restore activities of mitochondrial [4Fe–4S] enzymes in HeLa cells depleted for endogenous ISCA1. HeLa cells were transfected twice at a 3 days interval with scrambled control siRNAs (SCR), or with ISCA1-directed siRNAs (siISCA1) in combination with complementing (Compl.) plasmids encoding RNAi-resistant versions of wild-type or patient ISCA1 (smISCA1^wt or smISCA1^V10G, respectively). Additionally, cells were mock-transfected as specified. w/o indicates transfection of a non-related EGFP-encoding control plasmid. Analyses were performed after the second round of transfection, i.e. after a total of 6 days of ISCA1 depletion. (A) Total cell lysates were subjected to immunoblotting and analyzed for the steady-state protein levels of the indicated ISC proteins and β-actin as a loading control. (B) Immunoblot signals were quantified relative to β-actin levels and the ratio was normalized to mock-transfected control (Ctrl) cells. (C) The specific activity of RCCI was determined in a crude mitochondrial fraction which was obtained by mechanical cell disruption using the Potter–Elvehjem method. Specific activities of SDH (RCCII) and RCCIV were determined in mitochondria-containing membrane fractions prepared by digitonin-based cell fractionation. Values were normalized to mock-transfected control (Ctrl) cells. (D) LIAS activity was determined by immunoblotting for lipoylation of KGDH-E2 and PDH-E2. The latter protein (DLAT) was also visualized by immunostaining. (E) Immunoblot signals from (D) were quantified, and the protein per β-actin ratio was normalized to mock-transfected control (Ctrl) cells. (F) The specific activity of ACO2 was determined in samples from (C) and normalized to the values deduced from mock-transfected control (Ctrl) cells. The specific activity of citrate synthase (CS) was used as reference. (G) Total cell lysates were subjected to immunoblotting as in (A) and analyzed for the steady-state levels of the indicated proteins. (H) Immunoblot signals from (G) were quantified and the protein per β-actin ratios were normalized to mock-transfected control (Ctrl) cells. Representative blots are shown. All values are given as the mean ± SD (n = 4 to 5); * P < 0.05; ** P < 0.01; *** P < 0.001; ns, not significant; dashed line, 100%-value of mock-transfected control cells.

Analysis of RCCs activities in ISCA1-depleted HeLa cells revealed severe defects in the function of the [4Fe–4S] cluster-containing RCCI and RCCII as well as in RCCIV (6) and (Fig. 4C). These defects could be restored by smISCA1^wt almost to control levels. In contrast, complementation by the patient version smISCA1^V10G was markedly less efficient, resulting in up to 50% lower RCCs activities compared with smISCA1^wt-expressing cells. Accordingly, steady-state protein levels of RCCI, RCCII and RCCIV subunits were strongly diminished upon RNAi-mediated loss of ISCA1 (Supplementary Material, Fig. S1A, C and D), in line with previous observations (6,10). Expression of smISCA1^wt restored the RCCs subunit protein levels to those detected in control cells, whereas complementation by the smISCA1^V10G patient version was less pronounced and remained incomplete, particularly in the case of the RCCII Fe-S subunit SDHB and the RCCIV subunit COX2 (Supplementary Material, Fig. S1A and D). In contrast, F₁α/β subunits of RCCV were hardly affected by the treatments, rendering an all-embracing OXPHOS defect unlikely.

Similar results were obtained when we examined the function or steady-state protein levels of other mitochondrial [4Fe–4S] proteins. Analysis of LIAS function by probing the lipoylation of KGDH-E2 and PDH-E2 in ISCA1-deficient cells revealed a severe defect which was only incompletely complemented by the smISCA1^V10G mutant protein when compared with smISCA1^wt (Fig. 4D and E). Despite the fact that PDH-E2 (DLAT) steady-state protein levels in ISCA1 RNAi-treated cells were also slightly lower than in control cells, the PDH-E2 lipoylation normalized to DLAT protein was still lower in smISCA1^V10G-complemented compared with smISCA1^wt-complemented cells (Fig. 4E, right panel). Enzyme activities and steady-state protein levels of ACO2 were only mildly affected by short-term depletion of ISCA1 for 6 days (Fig. 4F–H) similar to previous observations (6). Owing to this mild depletion phenotype, both smISCA1^wt and smISCA1^V10G were able to revert the RNAi-mediated defects. An important but rarely analyzed mitochondrial [4Fe–4S] protein is the electron transfer flavoprotein dehydrogenase (ETFDH) that is critically involved in the transfer of electrons from fatty acid β-oxidation to ubiquinone of the mitochondrial respiratory chain (23). ISCA1-directed siRNAs induced a significant decrease in ETFDH steady-state levels (Fig. 4G and H), suggesting a destabilization of the protein owing to impaired maturation (24). As for ACO2 expression of both smISCA1^wt and smISCA1^V10G recovered wild-type ETFDH protein levels. Notably, neither RNAi-mediated ISCA1 depletion nor transfection of smISCA1^wt - and smISCA1^V10G-encoding plasmids significantly affected the [2Fe–2S] proteins UQCRFS1 (Rieske Fe-S protein) (Supplementary Material, Fig. S1A and D) or ferrochelatase (FECH) (Fig. 4G and H), supporting the specific function of ISCA1 in the maturation of mitochondrial [4Fe–4S] proteins. As expected from previous findings (6) probing for the maturation of nuclear or cytosolic Fe-S proteins by immunoblotting or enzyme activity assays did not reveal large defects (i.e. more than 20% decrease from control values) for several tested Fe-S proteins (DPYD, GPAT, NTHL1 and POLD1) and non-Fe-S proteins (Supplementary Material, Fig. S3A, B and E). Likewise, we did not observe major effects on proteins involved in the Fe-S cluster-related pathway of cellular iron regulation (IRP1-ACO1, IRP2, TFR1) (Supplementary Material, Fig. S3A and D). Two exceptions were weak effects on cytosolic aconitase activity (Supplementary Material, Fig. S3C) and on IOP1 steady-state protein levels (Supplementary Material, Fig. S3A and B), which likely react in a sensitive way to disturbances of the mitochondrial homeostasis as a consequence of ISCA1 deficiency (6).

Preventing proteolysis of ISCA1^V10G only partially improves the maturation of mitochondrial [4Fe–4S] proteins

Our phenotypical and functional analysis of ISCA1 patient and control culture samples had revealed a decreased steady-state level and activity of the smISCA1^V10G mutant protein. In order to test whether the cellular defects related to the ISCA1 disease mutation may be reverted by preventing potential proteolytic degradation of ISCA1^V10G, we treated control, ISCA1-depleted, as well as smISCA1^wt- and smISCA1^V10G-complemented cells with the protease inhibitor MG132 for 16 h prior to biochemical analyses (20). Immunoblotting demonstrated a substantial elevation of ISCA1^V10G protein close to control levels as well as a more than 4-fold overproduction of plasmid-expressed ISCA1^wt in the presence of MG132 (Fig. 5A and B), indicating that both wild-type and mutant ISCA1 undergo high proteolytic turnover in vivo, with ISCA1^V10G being particularly unstable even in the presence of the protease inhibitor. In contrast, other late-acting ISC components (IBA57, NFU1 and IND1) or mitochondrial MIA40 did not differ in their abundance (Fig. 5A and B;Supplementary Material, Fig. S4A and B). Despite the fact that the amount of ISCA1^V10G in the presence MG132 was nearly as high as that of endogenous ISCA1 in control cells, the mutant protein was not able to restore RCCI and RCCII (SDH) activities to wild-type levels (Fig. 5C). Accordingly, steady-state protein levels of several RCCI and RCCII subunits remained slightly lower in smISCA1^V10G compared with smISCA1^wt-complemented cells (Supplementary Material, Fig. S4A, C and D). Only RCCIV activity and subunit levels were benefitting from the MG132 treatment.

Figure 5.

Inhibiting the degradation of the ISCA1^V10G mutant protein does not fully restore its normal function. HeLa cells were transfected with siRNAs and ISCA1-encoding plasmids and samples were evaluated as in Figure 4, with the difference that cells were cultured in presence of 10 µM MG132 for 16 h prior to the final harvest. (A, B) Total cell lysates were subjected to immunoblotting and quantified for the steady-state protein levels of the indicated proteins. (C) The specific activities of RCCI, SDH (RCCII) and RCCIV were determined and normalized to mock-transfected control (Ctrl) cells. (D, E) LIAS activity was assessed by quantitating the lipoylation status of KGDH-E2 and PDH-E2, as well as by immunostaining for PDH-E2 (DLAT) protein. (F) The specific activities of ACO2 and CS were determined and normalized to the values deduced from mock-transfected control (Ctrl) cells. (G, H) Total cell lysates were subjected to immunoblotting as in (A) and quantified for the steady-state protein levels of the indicated proteins. Representative blots are shown. Staining for β-actin served as control (Supplementary Material, Fig. S5A and E). All values are given as the mean ± SD (n = 3–4); *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant; dashed line, 100%-value of mock-transfected control cells.

The effects of MG132 treatment were also tested for LIAS activity. Expression of smISCA1^wt and ISCA1^V10G complemented the lipoylation defect of both KGDH-E2 and PDH-E2 subunits to a statistically significant extent (Fig. 5D and E). Likewise, insignificant differences between smISCA1^wt- and smISCA1^V10G-complemented cells were detected for the activity and protein levels of ACO2 and ETFDH (Fig. 5F–H). Furthermore, probing for the maturation of extra-mitochondrial Fe-S proteins did not reveal any profound defects (Supplementary Material, Fig. S5A, B, D and E). Notably, the effect on cytosolic aconitase activity observed above was no longer detectable after MG132 treatment (cf. Supplementary Material, Figs S3C and S5C). Taken together, our results indicate that the decreased protein amount of ISCA1^V10G may partially but not fully explain the observed patient phenotype.

The p.V10G mutation present in the uncleaved presequence of ISCA1 does not affect correct targeting to the mitochondrial matrix

Since the destabilizing, disease-causing mutation is residing within the presumed mitochondrial presequence, we examined whether ISCA1^V10G was still correctly targeted to the mitochondrial matrix or at least partially remained in the cytosol where it may be prone to degradation. To this end, HeLa cells were transfected and treated with MG132 as above, and the subcellular localization of endogenous and ectopically expressed ISCA1 proteins was followed by immunoblotting subsequent to digitonin-based cell fractionation (6). Both endogenous and ectopically expressed ISCA1^wt and ISCA1^V10G were predominantly present in the mitochondria-containing membrane fraction which also contains mitochondrial NFU1, F1α/β and MIA40 (Fig. 6A). In contrast, the cytosolic fraction with IRP1 and α-tubulin as control proteins did not contain significant amounts of ISCA1. In order to test the sub-mitochondrial localization of smISCA1^wt and smISCA1^V10G, the membrane fractions were first subjected to treatment either in hypotonic buffer to rupture the mitochondrial outer membrane (MOM) or in detergent-containing solution (Triton X-100) to dissolve both mitochondrial membranes. Samples where then incubated with proteinase K (PK) to digest protease-accessible proteins. Immunoblotting revealed that, in contrast to cytosolic β-actin (a contaminating protein of these membrane fractions), endogenous ISCA1 and both plasmid-expressed smISCA1 protein versions were sensitive to PK only in presence of detergent (Fig. 6B, Lanes 6, 12, 18, 24). This behavior was similar to that of NFU1, another matrix-localized protein. In contrast, PK treatment of mitochondria after their exposure to hypotonic buffer resulted in the degradation of the mitochondrial intermembrane space protein MIA40 indicating efficient mitochondrial swelling and rupture of the MOM (Fig. 6B, Lanes 4, 10, 16, 22). We conclude from these results that all ISCA1 proteins including ISCA1^V10G were located predominantly in the mitochondrial matrix, excluding an efficient mis-targeting as a result of the V10G amino acid exchange.

Figure 6.

The p.V10G ISCA1 mutation within the uncleaved presequence interferes with import of ISCA1 into mitochondria. (A) HeLa cells were transfected with siRNAs and ISCA1-encoding plasmids prior to MG132 treatment as in Figure 5. Total cell lysates, cytosolic and membrane samples were obtained by digitonin-based cell fractionation and analyzed for the indicated proteins by immunostaining. (B) HeLa cells were transfected once, and treated with MG132 as in (A). Membrane fractions were subjected to proteinase K (PK*; non-specifically stained by the anti-NFU1 antiserum) digestion after treatment with hypotonic buffer (Hypo) or the detergent Triton X-100 (TX-100). PK sensitivity of indicated proteins was analyzed by immunostaining. (C) HeLa organellar material derived from digitonin-based cell fractionation was mixed with increasing amounts of in vitro translated [³⁵S] radiolabeled ISCA1^wt as indicated. Mixed (Lanes 3–5) and non-mixed (Lanes 1 and 6) samples were resolved by 16% tricine–SDS gel electrophoresis followed by blotting, autoradiography (left panel), and anti-ISCA1 immunostaining (right panel). MW; molecular mass markers. (D) Upper panel: representative autoradiograph of a mitochondrial import experiment using ³⁵S radiolabeled ISCA1^wt (Lanes 1–5) or ISCA1^V10G (Lanes 6–10). Radiolabeled proteins were incubated with mitochondria isolated from HeLa cells in the absence or presence of 1 μM CCCP (carbonyl cyanide m-chlorophenyl hydrazone, a protonophore of the mitochondrial inner membrane). Trypsin proteolysis was performed in the presence Triton X-100 where indicated. Samples were analyzed by SDS-PAGE and autoradiography. Lower panel: samples were subjected to immunostaining against NDUFA9 serving as loading control. (E) The ³⁵S radiolabeled ISCA1^wt or ISCA1^V10G proteins were produced in vitro in the RRL system, and detected by SDS-PAGE followed by autoradiography. The amounts shown represent half of the material incubated with mitochondria in (C).

In order to determine whether ISCA1 undergoes presequence processing upon mitochondrial import, we mixed a mitochondria-containing fraction with in vitro translated, ³⁵S-radiolabeled full-length ISCA1^wt, and analyzed ISCA1 proteins by immunoblotting and autoradiography (Fig. 6C;Supplementary Material, Fig. S6A–C). The ³⁵S-radiolabeled ISCA1^wt was detectable by both techniques and matched exactly the size of ISCA1 of the mitochondria-containing cell fraction. These results indicate that the presequence of human ISCA1 is not removed upon mitochondrial import. Hence, the mutation p.V10G is part of the mature ISCA1 protein.

The p.V10G mutation of ISCA1 affects the mitochondrial import efficiency

As shown above, at moderate expression levels mutant ISCA1^V10G appears to properly reside within the mitochondrial matrix (Fig. 6B). However, in some of our MG132 experiments we achieved over-expression of both ISCA1^wt and ISCA1^V10G (Supplementary Material, Fig. S7). In these cases, immunoblotting after cell fractionation revealed that a fraction of mutant ISCA1^V10G had escaped mitochondrial import and was present in the cytosolic fraction, even though the majority of ISCA1^wt was still found within the organelle (Supplementary Material, Fig. S7). To directly assess whether the p.V10G mutation of ISCA1 affects the mitochondrial import efficiency, we performed in vitro import assays by incubating freshly isolated mitochondria from HeLa cells with in vitro translated, ³⁵S-radiolabeled ISCA1^wt or ISCA1^V10G (Fig. 6D). The ³⁵S-radiolabeled protein products were checked by SDS-PAGE followed by autoradiography to ensure equal addition of in vitro translated ³⁵S-radiolabelled ISCA1^wt or ISCA1^V10G to the mitochondrial import mixture (Fig. 6E). Equal mitochondrial loading was verified by immunoblotting for the NDUFA9 subunit of RCCI (Fig. 6D, bottom panel). The efficient uptake of ³⁵S-ISCA1^wt into mitochondria became evident from its resistance to cleavage by added trypsin (Fig. 6D, Lanes 1 and 2), unless mitochondria were lysed by Triton X-100 treatment (Fig. 6D, Lane 3). Mitochondrial import of ³⁵S-ISCA1^wt was dependent on a mitochondrial membrane potential, as the uncoupler CCCP completely inhibited the translocation into the organelles (Fig. 6D, Lanes 4 and 5). Under these conditions, some radioactive ³⁵S-ISCA1^wt was still bound to mitochondria (Fig. 6D, Lane 4), yet this represents protein associated to the mitochondrial surface, as indicated by its complete trypsin sensitivity (Fig. 6D, Lane 5). In contrast, binding of the ³⁵S-ISCA1^V10G variant to mitochondria was severely lower than that of ISCA1^wt (Fig. 6D, Lane 6 versus 1) and, importantly, was fully sensitive to trypsin cleavage even in the absence of Triton X-100, indicating a complete failure of mitochondrial import (Fig. 6D, Lanes 7 and 8). Taken together, these results indicate that the p.V10G amino acid exchange of ISCA1 substantially interferes with its mitochondrial import, yet the mutant protein appeared to retain some residual in vivo import into the matrix.

Increasing mitochondrial ISCA1^V10G mutant protein levels only partially improves [4Fe–4S] protein maturation

We finally asked whether increased amounts of mitochondrial ISCA1^V10G may fully revert the ISCA1 depletion phenotypes or whether the mutation in the uncleaved presequence piece is of functional relevance. To this end, we enhanced the mitochondrial import efficiency of ISCA1 by fusing a potent targeting sequence (from Neurospora crassa subunit 9 of mitochondrial ATPase, Su9) to the N terminus of full-length smISCA1^V10G and smISCA1^wt. Expression of Su9-smISCA1^V10G and Su9-smISCA1^wt in ISCA1-depleted HeLa cells resulted in a proper mitochondrial targeting of the two proteins as indicated by their colocalization with F1α/β and MIA40 (Fig. 7A). The IRP1- and tubulin-containing cytosolic fraction did not contain any detectable amounts of ISCA1. The electrophoretic migration of both mature mitochondrial Su9-smISCA1 fusion proteins was slightly slower than that of endogenous ISCA1, consistent with the presence of four residues remaining from the added Su9 prepiece. Immunoblotting of total lysate and mitochondrial fractions revealed an additional ISCA1 band with higher molecular mass consistent with an uncleaved Su9 presequence (calculated mass 7 kDa). The fusion protein was present exclusively in samples of Su9-smISCA1-complemented cells. Intriguingly, the relative amount of the non-processed p.V10G mutant protein was similar to that of wild-type Su9-ISCA1^wt (121 ±19%, n = 3, P > 0.1). This was in contrast to the substantially lower levels of mature Su9-smISCA1^V10G compared with wild-type Su9-smISCA1 (Fig. 7A and C). Together, these findings indicate a preferential degradation of the mutant ISCA1^V10G protein in the mitochondrial matrix. Nevertheless, the mitochondrial abundance of processed Su9-smISCA1^V10G was about two times as high as endogenous ISCA1 in control cells, whereas IBA57 and NFU1 levels were unchanged (Fig. 7B and C). The elevated amount of Su9-smISCA1^V10G was still unable to restore the activity of RCCII to levels of Su9-smISCA1^wt-complemented cells (Fig. 7D). Likewise, the ETFDH protein levels (Fig. 7E and F) and the LIAS activity in lipoylation of KGDH-E2 (Fig. 7G and H) were only incompletely reverted by expression of mutant Su9-smISCA1^V10G. In contrast, the activities of RCCI, RCCIV and ACO2 were increased with similar efficiency by both Su9-smISCA1 proteins (Fig. 7D–F, I), reminiscent of the findings in the presence of MG132 treatment. Together, these findings suggest that even elevated mitochondrial ISCA1^V10G levels only partially restore the enzyme activities diminished by ISCA1 depletion (Fig. 7D–H). In summary, the combined effects of the severely impaired mitochondrial import efficiency, the decreased stability, and the impaired [4Fe–4S] protein maturation function of ISCA1^V10G comprehensively explain the functional deficiency of ISCA1 mutant cells in mitochondrial [4Fe–4S] protein biogenesis.

Figure 7.

Promoting the mitochondrial import of ISCA1^V10G mutant protein cannot completely restore the defects of RNAi-mediated ISCA1 depletion. HeLa cells were transfected with siRNAs and Su9-ISCA1-encoding plasmids as indicated, and samples were evaluated as in Figures 4 and 6. (A, C) Total cell lysates, cytosolic and membrane samples were obtained by digitonin-based cell fractionation and analyzed for the indicated proteins by immunoblotting. (B, C) Total cell lysates were subjected to immunoblotting and quantified for the levels of the indicated proteins. (D) The specific activities of RCCI, SDH (RCCII) and RCCIV were determined and normalized to mock-transfected control (Ctrl) cells. (E, F) Total cell lysates were subjected to immunoblotting as in (B) and quantified for the levels of the indicated proteins. (G, H) LIAS activity was assessed by quantitating the lipoylation status of KGDH-E2 and PDH-E2, as well as by immunostaining for PDH-E2 (DLAT) protein. (I) The specific activities of ACO2 and CS were determined and normalized to the values deduced from mock-transfected control (Ctrl) cells. Representative blots are shown. Staining for β-actin served as control. All values are given as the mean ± SD (n = 3); *P < 0.05; **P < 0.01; ***P < 0.001; ns, not significant; dashed line, 100%-value of mock-transfected control cells.

Discussion

Here, we report on an Italian patient from non-consanguineous parents presenting with early-onset leukodystrophy, spastic ataxia, feeding difficulties and lactic acidosis, owing to a biallelic homozygous mutation in the ISCA1 gene. ISCA1 belongs to the A-type ISC proteins and functions as a late-acting component of the mitochondrial Fe-S protein assembly machinery. Recent studies have shown that ISCA1 and ISCA2 in both yeast and human cells form a hetero-complex that is involved in the generation of [4Fe–4S] but not [2Fe–2S] proteins (6,8). Depletion experiments in yeast and human cell culture suggest that the two proteins tightly cooperate in their function, since their ablation is associated with virtually identical phenotypes resulting from a global mitochondrial [4Fe–4S] protein deficiency (6,8). Knock-downs of the two ISCA proteins in mouse skeletal muscle showed that ISCA1 can also act independently of ISCA2 in [4Fe–4S] protein assembly by forming homo-dimers (10). Both ISCA proteins functionally interact with the late ISC protein IBA57 in yeast (8), while in human cells IBA57 appears to have a binding preference for ISCA2 (10). While the precise function of IBA57 is hitherto unknown, its depletion elicits strikingly similar defects as that of the two ISCA proteins in cell culture (6) suggesting that the three proteins may function at the same step of biogenesis.

Several patients carrying mutations in IBA57 have been intensely investigated and present with phenotypes ranging from a mild form of hereditary spastic paraplegia to a severe myopathy and encephalopathy combined with a respiratory insufficiency and hyperglycinemia (20,21,25–28). Biochemically, the IBA57 functional impairment in these patients leads to a defect of several mitochondrial [4Fe–4S] enzymes including RCCI, RCCII and LIAS, the latter causing diminished lipoylation and consequently decreased activities of the four lipoic acid-dependent enzymes PDHc, KGDHc, BCKGDHc and glycine cleavage system (6,21). Individuals with mutations in ISCA2 show a similar clinical phenotype as IBA57 patients and present with leukodystrophy and neuroregression (15,29,30). The first mutation (p.Gly77Ser) reported in ISCA2 (15), considered later as founder mutation (29), has been very recently found in two additional unrelated cases (30). This mutation diminished the mitochondrial DNA content, the mitochondrial membrane potential, the ATP production, and the activities of RCCII + RCCIV. Moreover, it has been observed that this mutation produces a defect in the functioning of [4Fe–4S] but not [2Fe–2S] proteins (30). Two additional unrelated patients carrying novel homozygous ISCA2 mutations (p.Leu52Phe or p.Arg105Gly) displayed different phenotypes, with the first patient showing milder defects from a clinical and biochemical point of view, and the second patient presenting with leukodystrophy and biochemical impairment (RCCI and CII reduction, lipoylation and mitochondrial aconitase defects) (31). A recent publication on two unrelated patients presenting with leukodystrophy has provided evidence for a mutation in the ISCA1 gene, yet no detailed molecular and biochemical description of the patient phenotype has been performed (22).

Our comprehensive clinical and biochemical characterization of a new ISCA1 patient shows conspicuous similarities with well documented IBA57 cases and the severe case of ISCA2 (30), as well with the recently published ISCA1 patients from two Indian families both sharing the same homozygous genotype (22). Comparing our patient with the two Indian ISCA1-related leukodystrophy sibships, the disease progressed slower. In fact, the affected Indian children all died within the age of 5 years, while our patient survived until the age of 11 years. Moreover, the MRI abnormalities were milder in our patient showing a pattern of a vacuolating leukodystrophy without cortical abnormalities that are reported in the two Indian sibships (22). The ISCA1 patient mutation markedly affected the maturation and the activity of mitochondrial [4Fe–4S] but not of [2Fe–2S] proteins. The most severe effects were observed for RCCI and RCCII, as well as for LIAS as indicated by the decreased activity and stability of the lipoylated enzymes in patient fibroblasts. In contrast, RCCIII with its [2Fe–2S] cluster was unaffected. Interestingly, the RCCs activities measured on muscle and brain homogenate also displayed a diminution of RCCIV, as previously observed in IBA57 patients and in cultured HeLa or yeast cells depleted of these ISC assembly factors (ISCA1, ISCA2 or IBA57) (6,20). In humans, this effect is probably the (indirect) result of respiratory super-complex destabilization caused by, e.g. RCCI defects (6). Our patient cell-derived findings support and extend the model of a decisive and specific role of ISCA1 in mitochondrial [4Fe–4S] protein biogenesis (6,10).

To further confirm and better understand the molecular consequences of the ISCA1 patient mutation, we took advantage of cultured HeLa cells depleted of ISCA1 and expressing wild-type or mutant versions of the protein. While the profound defects in mitochondrial [4Fe–4S] proteins upon ISCA1 depletion could be rescued by expression of wild-type ISCA1, the mutant ISCA1^V10G protein was much less effective, documenting the pathogenic character of the p.V10G mutation. ISCA1^V10G was correctly localized to the mitochondrial matrix, yet was present at low amounts. Since the p.V10G amino acid exchange is located near the N-terminus of ISCA1 and is predicted to be part of the mitochondrial presequence, we reasoned that the low amounts of ISCA1^V10G observed in both patient fibroblasts and rescued HeLa cells were caused by its inefficient import into the mitochondrial matrix. This view could be convincingly supported by an in vitro import assay showing that the ISCA1^V10G was unable to enter the mitochondria. Even though we have no direct evidence for an ISCA1^V10G import defect in living cells, the small amounts of mitochondria-associated ISCA1^V10G compared with wild-type protein are compatible with such a view. Moreover, inhibiting proteolysis by addition of the protease inhibitor MG132 resulted in small amounts of ISCA1^V10G in the cytosol indirectly indicating its impaired mitochondrial import. MG132 addition resulted in increased amounts of both wild-type and mutant ISCA1 protein, documenting a high intra-mitochondrial turnover of both ISCA1 proteins. ISCA1^V10G was still present at lower levels compared with plasmid-expressed wild-type ISCA1, indicating a particularly high sensitivity to proteolytic degradation. Although the proteolysis inhibition allowed ISCA1^V10G to almost reach levels of endogenous ISCA1 in control cells, yet resulted in only a partial recovery of the [4Fe–4S] enzyme defect. This suggests that the combination of decreased mitochondrial import and increased degradation of ISCA1^V10G does not fully explain the pathogenic character of the mutation. Further support for this notion comes from the experiments promoting mitochondrial import by a strong presequence (Su9; Fig. 7). This strategy yielded 2-fold higher levels of matrix-localized ISCA1^V10G compared with native ISCA1 in control cells, yet we still observed [4Fe–4S] protein maturation defects, e.g. on RCCII activity and KGDH lipoylation. These defects can only be attributed to a functionally impaired ISCA1. Since the altered residue p.V10G is part of the uncleaved presequence and hence present in the mature protein, the N-terminal piece of ISCA1 plays a crucial functional role that has not been appreciated hitherto.

The functional consequences of an ISCA1 deficiency are mainly restricted to mitochondria. The only exceptions were ACO1-IRP1 and IOP1 whose analysis pointed to a slightly decreased maturation efficiency. Similar effects on ACO1-IRP1 have been observed also for IBA57- and ISCA2-depleted HeLa cells and for the cytosolic Fe-S protein Leu1 in yeast, an aconitase-like protein (6,8). However, since other cytosolic-nuclear Fe-S proteins were hardly affected, these moderate effects seem to be an indirect consequence of the mitochondrial dysfunction. The effect on ACO1-IRP1 was still too weak to elicit any secondary alteration of the cellular iron homeostasis as indicated by unchanged levels of IRP2 and transferrin receptor, two critical members of cellular iron metabolism. The lack of any clinical signs of a disturbed iron homeostasis in ISCA1, ISCA2 and IBA57 patients further supports the view of a weak and indirect effect on ACO1-IRP1.

As mentioned above, defects in biogenesis of mitochondrial Fe-S proteins give rise to a wide range of clinical phenotypes, including myopathy, anemia, encephalopathy, lactic acidosis, cardiomyopathy, non-ketotic hyperglycinemia and pulmonary hypertension. Nevertheless, vacuolating lesions of the white matter seem to be a frequent feature resulting from defects of the late-acting ISC components such as ISCA1 (this work), ISCA2 (15), IBA57 (21,26) as well as BOLA3 (32), NFU1 (17,33,34) and IND1 (35). The latter factors act after the ISCA-IBA57 proteins and facilitate the insertion of the [4Fe–4S] cluster into the recipient apoproteins (1,14,36). Accordingly, our ISCA1 patient shows an MRI pattern of vacuolating leukodystrophy with a distribution of abnormalities that do not particularly differ from what has been described for other MMDS. This feature is characterized by symmetrical patchy areas of the white matter that aggregate in larger cavitating areas, until it affects all the structures of the white matter including the internal capsule and the corpus callosum. This form of leukodystrophy, defined as progressive cavitating leukoencephalopathy, was initially described by Naidu and coworkers in 19 children with undefined molecular diagnosis (37). In the last years, many mitochondrial genes have been associated to leukodystrophy (34,38–41) and some of them are well recognizable by MRI patterns, particularly in those related to defects of some aminoacyl tRNA synthetases (42–44). Although mutations in the late-acting components of the Fe-S cluster biogenesis are not distinctly recognizable by MRI, a screening for mutations in IBA57, ISCA1, ISCA2, NFU1 and BOLA3 genes in patients affected by cavitating leukodystrophy with combined or isolated defects of RCCI and RCCII should always be performed.

Our findings on ISCA1, combined with earlier publications on IBA57 and ISCA2, suggest that a dysfunction of these late-acting ISC factors in patients elicits strikingly similar clinical and biochemical phenotypes. These features can now potentially be exploited for rapid diagnosis of new disease candidates by measuring, e.g. the lipoylation status of proteins together with respiratory activities. Collectively, our results clearly show that the ISCA1 p.V10G mutation located within the uncleaved mitochondrial presequence decreased mitochondrial import of ISCA1^V10G as well as diminished its protein stability and biochemical function in the matrix. These combined effects fully explain the neurological phenotype and the maturation defects of mitochondrial [4Fe–4S] proteins observed in our patient.

Materials and Methods

Standard protocol approvals, registrations and patients consents

The study was approved by the Ethical Committees of the ‘Bambino Gesù’ Children’s Hospital, Rome, Italy, in agreement with the Declaration of Helsinki. Informed consent was signed by the parents of the patient.

Tissue culture and transfection

Human fibroblasts were obtained from a diagnostic skin biopsy and grown in high-glucose DMEM medium supplemented with 10% fetal bovine serum, 4.5 g/l glucose, and 50 μg/ml uridine. HeLa cells were cultured in high-glucose DMEM supplemented with 7.5% fetal calf serum, 2 mM stabilized glutamine, and 1% Pen-Strep (50 U/ml penicillin and 50 µg/ml streptomycin). Transfection by electroporation was carried out two times in a 3-day interval as described (6,20). Briefly, 6.5 × 10⁶ HeLa cells were resuspended in 265 µl of transfection buffer, supplemented with 3 nmol of each of three siRNA oligonucleotides, 5 µg of complementing or control plasmids and 3.75 µg of pVA1 plasmid in order to improve ectopic gene expression and cell recovery after electroporation. ISCA1-directed siRNAs and RNAi-resistant (silently mutated) smISCA1 were adopted from (6). The c.29T>G exchange resulting in the p.V10G exchange was introduced by PCR-based site directed mutagenesis using a common ‘round-the-horn’ approach (6). In order to improve mitochondrial targeting of ISCA1, the coding sequence of N. crassa subunit 9 of mitochondrial ATPase (Su9) (45) was cloned in front of the ISCA1 open reading frame. In some experiments the protease inhibitor (S)-MG132 (Cayman) was added to the culture medium at a concentration of 10 µM 16 h prior to cell harvest. MG132 was prepared freshly by dissolving in DMSO at a concentration of 10 mM under oxygen-free conditions and stored frozen.

Biochemical and protein studies

RCCs and citrate synthase activities were assayed on muscle homogenate using a reported spectrophotometric method (46). Complex V activity (in the direction of ATP synthesis) was measured in fibroblasts mitochondria of the patient and age-matched controls, using either succinate, malate or pyruvate + malate as substrates (47). The KGDHc, PDHc and (BCKDHc enzymatic activities were assayed essentially as previously described) (48,49).

For BNGE fibroblasts mitochondria were processed as described elsewhere (50) and 40 μg of proteins were loaded on linear 5–12% non-denaturing gradient gel for first dimension. For SDS-PAGE, 35–45 μg of fibroblasts mitochondria were loaded in a 12% denaturing gel. Small proteins were resolved by reducing 16% tricine–SDS gel electrophoresis (51). WB of either BNGE or SDS-PAGE was achieved by transferring proteins onto polyvinylidenedifluoride or nitrocellulose membranes and probed with specific antibodies. Specific bands were detected using Lite Ablot Extend Long Lasting Chemiluminescent Substrate [Euroclone, Pero (MI), Italy]. Densitometry analysis was performed using Quantity One software (BioRad, Hercules, CA, USA). Digitonin-based fractionation of HeLa cells, preparation of crude mitochondria, as well as determination enzyme activities by spectrophotometric and immunoblotting approaches have been performed according to published methods (20,52,53).

Sub-mitochondrial localization of proteins was performed according to (54) by adding 15 µl of mitochondrial fractions (protein concentration 5 µg/µl) to 135 µl iso-osmotic buffer (25 mM Tris/HCl, 250 mM sucrose, 1.5 mM MgCl₂, pH 7.4), 300 µl hypo-osmotic buffer (5 mM HEPES, pH 7.4), or 135 µl iso-osmotic buffer with 0.2% Triton X-100. Proteinase K was added to a final concentration of 0.1 µg/µl and samples were incubated for 30 min at 37°C. Finally, proteins were precipitated by 10% trichloroactic acid and solubilized in Laemmli buffer for subsequent immunoblotting. Protein determination was performed by the BCA assay.

Antibodies

For fibroblasts experiments mouse monoclonal antibodies were obtained from MitoScience (Eugene, OR, USA): complex I—NDUFS1; complex II—SDHA, SDHB; complex III—UQCRC2; complex IV—COXIV; complex V—F1β; porin (VDAC) and PDH cocktail. Polyclonal rabbit TOMM20 was purchased from Santa Cruz Biotechnology, Inc.; anti-GAPDH from Sigma-Aldrich (Saint Louis, Missouri, USA); anti-lipoic acid and anti-LIAS were commercially available from Abcam (Cambridge Science Park, UK); anti-ACO2 from Novus Biologicals (Abingdon Science Park, UK). Rabbit monoclonal anti-ACO1 was from Abcam whereas mouse monoclonal anti-β-actin was from Sigma-Aldrich. Polyclonal anti-ISCA2 has been introduced in (6).

For Hela cell immunoblotting mouse monoclonal antibodies, NDUFA13, NDUFB4, NDUFB6 were obtained from MitoScience. Purified polyclonal rabbit antibodies directed against UQCRFS1, UQCRC2, COX2, COX6a/b and F1α/β were a kind gift of H. Schägger and I. Wittig (Frankfurt, Germany). Mouse monoclonal anti-β-actin (Santa Cruz), anti-DLAT (Cell Signaling Technology, 4A4-B6-C10), anti-GAPDH (Calbiochem), anti-IRP2 (clone 7H6, Santa Cruz), anti-NTHL1 (Santa Cruz), anti-TFR1 (clone H68.4, ThermoFisher) and anti α-tubulin (clone DM1α, Sigma Aldrich) as well as rabbit polyclonal anti-lipoic acid and anti-LIAS (Abcam), anti-ACO1 (Invitrogen), anti-ACO2 (Novus Biologicals), anti-DPYD (Santa Cruz) and anti-POLD1 (PTG lab) antibodies were commercially available. Purified rabbit anti-GPAT was kindly provided by H. Puccio (Illkirch, France), and mouse anti-IRP1 was a kind donation of R. Eisenstein (Madison, USA). Anti-IOP1 antiserum was raised in rabbits using purified proteins produced in E. coli.

The following antibodies were used for common experiments: NDUFA9 (MitoScience), anti-ISCA1 and anti-IBA57 (antisera were raised in rabbits using purified proteins produced in E. coli).

Mutational analysis

Genomic DNA was isolated from blood and cultured skin fibroblasts using QIAamp DNA mini kit (QIAGEN, Valencia, CA, USA). Patient DNA was subjected to targeted resequencing, outsourced at the BGI-Shenzhen (BGI-Shenzhen, Shenzhen, China). A custom probe library was used for targeted enrichment (Agilent SureSelectXT Custom Kit) designed to capture coding exons and flanking intronic stretches (20 nts) of 1381 genes encoding for mitochondrial proteins (Mitoexome) (55), followed by deep sequencing using Illumina Hiseq technology, providing 255× effective mean depth. Validation of the ISCA1 variant and segregation along the family was performed by Sanger Sequencing, using BigDye chemistry 3.1 and run on an ABI 3130XL automatic sequencer (Applied Biosystems, Life Technologies).

In vitro transcription and translation of radiolabeled ³⁵S methionine/cysteine ISCA1 and ISCA1-V10G proteins

Full-length human ISCA1^wt and ISCA1^V10G cDNAs were generated by reverse transcriptase-PCR, using RNA extracted from primary skin fibroblasts isolated from control subjects and patient, respectively. The ISCA1^wt and ISCA1^V10G cDNAs were subcloned in pCDNA3.1(+) vector with the T7 promoter. Plasmid constructions were confirmed by DNA sequencing. In vitro transcription and translation were performed in Rabbit Reticulocyte Lysate System (RRL) (Promega Biotech, Madison, WI, USA) as recommended by the supplier. One microgram of ISCA1^wt or ISCA1^V10G construct was added to 50 μl Promega standard mixture, containing T7 RNA polymerase and a standard amino acid mixture with ³⁵S Met/Cys (20 μCi). Incubation was at 30°C for 90 min. For analysis of ISCA1 presequence processing, hemoglobin present in the rabbit reticulocyte lysate was removed by an ammonium sulfate precipitation approach (56).

Import of ³⁵S met/cys-labeled ISCA1 and V10G proteins in mitochondria isolated from HeLa cell cultures

Cells were collected by centrifugation at 500×g for 3 min and homogenized in ice-cold homogenization buffer (250 mM sucrose, 1 mM EDTA and 10 mM Tris−HCl, pH 7.4). Nuclei and cellular debris were pelleted by centrifugation at 3000×g for 8 min. The supernatant was centrifuged at 10 000×g for 12 min, resulting in a crude mitochondrial pellet. Mitochondria were immediately employed for import experiments. RRL translation products were incubated with freshly isolated mitochondria in BSA buffer (20 mM HEPES–KOH, pH 7.4, 250 mM sucrose, 80 mM potassium acetate, 5 mM magnesium acetate, 3% BSA) plus 10 mM sodium succinate and 2 mM ATP for 60 min at 37°C. Mitochondria were spin down from the import mixture, resuspended in BSA buffer and, where indicated, treated with trypsin (0.5 mg per 50 mg mitochondrial proteins, 35 min on ice) in the absence or in the presence of 0.2% Triton X-100. Proteins were then analyzed by SDS-PAGE and autoradiography. Aliquots of all samples were subjected to immunoblotting against the NDUFA9 subunit of Complex I.

Statistical analysis

Data are presented as mean ± SD. In patient-related analyses Student's t-test was used for the determination of statistical significance. Tissue culture RNAi and complementation experiments were performed and examined in a batch-dependent fashion. Datasets were analyzed by one-way repeated measures ANOVA followed by Dunnetts post hoc test, with smISCA1^wt data serving as control in order to individually determine the significance levels of ISCA1 depletion phenotypes as well as of smISCA1^wt and smISCA1^V10G complementation phenotypes. # indicates a statistical trend (P < 0.1). Significance levels: *P < 0.05, **P < 0.01 and ***P < 0.001.

Supplementary Material

Supplementary Material is available at HMG online.

Acknowledgements

E.B., R.C. and A.T. acknowledge a grant from the Italian Ministry of Health (Ricerca Corrente 2016–2017). R.L. was supported by Deutsche Forschungsgemeinschaft (LI 415/6, SPP 1710, and SPP 1927).

Conflict of Interest statement. None declared.

Author notes

Alessandra Torraco and Oliver Stehling authors contributed equally to this work.

Enrico Bertini, Roland Lill and Rosalba Carrozzo authors contributed equally as senior authors.