CIA30 and complex I assembly

CIA30 has been implicated in the assembly of complex I in N. crassa by virtue of its association with a stalled assembly intermediate (Kuffner et al, 1998). We asked whether the human CIA30 ortholog associates with complex I subunits in mitochondria. Co-immunoprecipitations of mitochondrial lysates were performed using CIA30 antibodies coupled to protein A-Sepharose (Figure 2). Antibodies from preimmune sera coupled to protein A-Sepharose served as a negative control. Specifically bound proteins were eluted from the beads and subjected to SDS–PAGE followed by Western blot analysis using antibodies against a number of mitochondrial proteins. A sample of the unbound mitochondrial fraction (representing 2% of total) was also analyzed. As can be seen, CIA30 had been completely extracted from the mitochondrial lysate (Figure 2, top panel, compare lanes 2 and 4) while antibodies from the preimmune sera did not pull down CIA30 (Figure 2, top panel, lanes 1 and 3). We next determined the presence of various complex I subunits including the mtDNA-encoded subunit ND1 and the nuclear-encoded subunits NDUFA9, NDUFS3, NDUFS7, NDUFB6, NDUFS5 and NDUFA6. All subunits co-immunoprecipitated with CIA30, except NDUFS5 (Figure 2), which appears to assemble directly into complex I (Ugalde et al, 2004b). Little, if any, protein was detected when preimmune antibodies were used in the co-immunoprecipitations. Subunits from complexes III (core 1) and II (70 kDa) were not immunoprecipitated with CIA30, suggesting that binding is specific for complex I subunits. Unlike CIA30 however, the complex I subunits were not significantly immunodepleted from the mitochondrial extracts since they were all found in the unbound fraction (Figure 2, lane 4). This suggests that CIA30 associates with complex I subunits found only in intermediate subcomplexes and not those within mature complex I. It has recently been shown that NDUFS5 (15 kDa subunit) is not found in any detectable assembly intermediate and may be incorporated directly into the holocomplex at a late stage (Ugalde et al, 2004b). The finding that NDUFS5 does not stably interact with CIA30 in our experiments supports these findings.

Open in a separate window

Figure 2

Co-immunoprecipitation of complex I subunits with CIA30. Mitochondria from HEK-293T cells were solubilized in 1% (w/v) Triton X-100 before immunoprecipitation with antibodies against CIA30 or preimmune serum coupled to protein-A Sepharose. The eluted fraction and a sample of the unbound fraction (2% of total) were then applied to Tris–Tricine SDS–PAGE followed by Western transfer and immunodecoration with antibodies as indicated.

We have recently found that newly translated mtDNA-encoded complex I subunits form detectable assembly intermediate complexes at ~460 and ~830 kDa on BN-PAGE (Lazarou et al, 2007). The latter complex is most likely the same as that found by Ogilvie et al (2005). After longer incubation times, these assembly intermediates disappear and the subunits are integrated into the fully assembled complex I. We determined whether CIA30 might be associated with intermediate complexes. MtDNA-encoded subunits were specifically radiolabeled in cultured human lymphoblasts for 2 h following a chloramphenicol pretreatment for 16 h. Mitochondria were then solubilized in Triton X-100 and subjected to immunoprecipitations using antibodies against CIA30 or from preimmune serum. The protein complexes not bound to these antibodies were analyzed by BN-PAGE. The preimmune sample shows the presence of a number of respiratory chain complexes containing newly incorporated radiolabeled mtDNA-encoded subunits (Figure 3A, lane 1). The most prominent species is complex V while complexes III, IV and a complex IV intermediate species labeled ‘CIV_i' can also be seen (McKenzie et al, 2007). In addition, the ~460 and ~830 kDa intermediate complexes containing complex I subunits were present in the unbound fraction following incubation with the preimmune sera (Figure 3A, lane 1). However, these intermediates were specifically depleted following incubation with immobilized CIA30 antibodies (Figure 3A, lane 2). These results indicate that CIA30 associates with these early subcomplexes that contain newly translated mtDNA-encoded subunits. The same radiolabeled samples were subsequently analyzed for the presence of CIA30 complexes using Western blotting (Figure 3A, lanes 3 and 4). With Triton X-100 solubilization, the major CIA30 complexes are at ~460 kDa (running as a doublet), with minor complexes at ~600 and ~830 kDa. These complexes were still present in the unbound fraction following incubation with preimmune antibodies, but they were specifically depleted with the CIA30 antibodies (Figure 3, compare lanes 3 and 4). Subsequent probing for complex I revealed that depletion of CIA30 had no effect on steady-state complex I levels, consistent with CIA30 not associating with the mature holoenzyme (Figure 3, lane 6). The levels of complex II were also assessed as a negative control to demonstrate the specificity of the immunodepletion and to serve as a loading control.

Open in a separate window

Figure 3

mtDNA-encoded complex I (CI) subunits associate with CIA30 at early stages in their assembly. (A) BN-PAGE analysis of mitochondria containing radiolabeled mtDNA-encoded subunits (0 h chase) immunodepleted with the αCIA30 antibody (lanes 2, 4 and 6). Preimmune serum was also used for control depletion (lanes 1, 3 and 5). Protein complexes containing radiolabeled mtDNA-encoded subunits were detected by autoradiography (lanes 1 and 2). After this, the lanes were subjected to Western blot analysis using antibodies against CIA30 (lanes 3 and 4) followed by antibodies against CI and complex II (CII) (lanes 5 and 6). (B) After radiolabeling, cells were chased for various times as indicated. Mitochondria were isolated, solubilized in Triton X-100 and incubated with protein A-Sepharose coupled to CIA30 antibodies or preimmune sera. Bound proteins were analyzed by SDS–PAGE and phosphorimaging (lanes 2–7). A labeled subunit profile (control) is shown in lane 1 and represents 16.5% of material used in each co-immunoprecipitation. The mtDNA-encoded subunits are indicated.

Our results suggest that CIA30 interacts with newly synthesized mtDNA subunits in complex I intermediates but not with the mature holoenzyme. To determine which mtDNA-encoded subunits associate with CIA30, a co-immunoprecipitation analysis was again undertaken. Mitochondrial extracts were used from cells that had been chased for various times after mtDNA subunit labeling. In this case, radiolabeled mtDNA-encoded subunits are present in intermediate assemblies at early time points while at later times (e.g. 24 h—see Figure 4C and McKenzie et al, 2007), they will have largely assembled into the complex I holoenzyme. Immunoprecipitated radiolabeled proteins were separated on SDS–PAGE and detected by phosphorimage analysis. Owing to the lack of available antibodies, only the ND1 subunit could be definitively identified based on its exact migration with endogenous ND1 (data not shown). Nevertheless, based on the mobilities of other subunits (COX1, 2, 3, cytochrome b, ATP6, ATP8 and ND5) and previous assignments (Chomyn, 1996), we could also confidently assign spots corresponding to subunits ND2 and ND3. At 0 and 3 h chase, the complex I subunits corresponding to ND1, ND2 and ND3 were co-immunoprecipitated with CIA30 (Figure 3B, lanes 3 and 5). The subunit corresponding to the mobility of ND5 was not specifically co-immunoprecipitated while ND4, ND4L and ND6 were not detected. The ATP8 subunit of complex V was also seen in the co-immunoprecipitates, but unlike the complex I subunits it was not specifically enriched against the total sample (lane 1), suggesting that it may represent a contamination. By 24 h chase, the radiolabeled subunits were no longer co-precipitated with CIA30 (Figure 3B, lane 7), consistent with these subunits integrating into the mature holoenzyme. Our results indicate that at early stages of complex I assembly, CIA30 is found in complexes of ~460 and ~830 kDa that contain (among others) the mtDNA-encoded subunits ND1, ND2 and ND3.

Open in a separate window

Figure 4

Identification of a complex I-deficient patient cell line with decreased CIA30. (A) Mitochondria isolated from fibroblasts of control (C) and patients (P1–P5) with complex I defects were applied to SDS–PAGE, Western blotted and probed with antibodies against CIA30, Tom40 and mtHsp70 (left panel). Patients had fibroblast complex I activities ranging from 10 to 50% of controls with unknown genetic cause (P2) or with two pathogenic mutations in the complex I subunit genes NDUFV1 (P3), NDUFS4 (P4) or NDUFS6 (P5). The steady-state levels of CIA30 in patient 1 (P1) were quantified against Tom40. Data are mean±s.e.m., n=3 (right panel). (B) Steady-state levels of complex I from patient 1 (P1) fibroblasts were analyzed by BN-PAGE and immunoblotting using antibodies against complex I (CI) and complex II (CII; left panel). SC=complex I/complex III₂ supercomplex. The steady-state levels of complex I were quantified against complex II. Data are mean±s.e.m., n=3 (right panel). (C) Mitochondrial extracts from control (C) and patient 1 (P1) fibroblasts were subjected to SDS–PAGE and Western blot analysis against complex I subunits and the complex II 70 kDa subunit (as control). (D) Respiratory chain enzyme activities in patient 1 (P1) lymphoblasts were measured and standardized against the mitochondrial marker enzyme citrate synthase and shown as a percentage of the control mean. Vertical lines represent the observed range for 6 control lymphoblast cell lines.

Identification of a complex I-deficient patient cell line with reduced levels of CIA30

A previous genetic screen of 13 patients with complex I deficiency but containing no known mutation in genes encoding any of its subunits, did not reveal any mutations in the NDUFAF1 gene encoding CIA30 (Janssen et al, 2002). Since this screening methodology did not focus on steady-state CIA30 protein levels, we undertook a Western blot analysis of fibroblast cell lines from patients diagnosed with enzymatically deficient complex I (Figure 4A, left panel). Although CIA30 varied in its levels between control and patient cell lines, it was severely reduced in patient 1 (Figure 4A), who had presented with cardioencephalomyopathy. Antibodies against mtHsp70 and Tom40 served as loading controls. Quantification against Tom40 revealed that the levels of CIA30 were ~10% compared to control (Figure 4A, right panel). CIA30 levels in patient 1 lymphoblasts were also decreased to ~10% of control values (data not shown). The level of steady-state complex I in patient 1 fibroblasts, as measured on BN-PAGE, was also reduced to ~30% of control values (Figure 4B). SDS–PAGE and Western blot analysis of mitochondrial extracts from patient 1 fibroblasts showed that the levels of all five complex I subunits tested were reduced compared to control mitochondria (Figure 4C). Consistent with the reduction of complex I steady-state levels, enzymatic analysis of patient 1 lymphoblasts showed a complex I deficiency of 36% residual activity relative to the marker enzymes citrate synthase or complex II. The activities of complexes II–IV were all normal (Figure 4D). Like lymphoblasts, patient 1 fibroblasts also showed a specific reduction in complex I activity (see Supplementary Figure 1).

Complex I assembly defects are found in patient 1 cells

The decreased levels of complex I and its subunits in patient 1 cells may be due to a defect in subunit assembly and/or increased turnover. To test this, we analyzed the assembly of newly synthesized mtDNA-encoded subunits into complex I. As can be seen (Figure 5A), at early chase times the ~830 kDa intermediate (marked ^*) is detectable in control fibroblasts (lanes 1 and 2), but is not visible in patient 1 fibroblasts (lanes 5 and 6). The ~460 kDa band is also reduced in patient 1 fibroblasts, although a faint band is visible at ~400 kDa (Figure 5A, lanes 5 and 6). By 24 h chase, mature complex I is detectable in control fibroblasts (Figure 5A, lane 4), but is not present in patient 1 fibroblasts (lane 8). The assembly of mtDNA-encoded subunits into the other respiratory chain complexes is not altered, demonstrating that the defect lies specifically in complex I assembly.

Open in a separate window

Figure 5

Complex I assembly is impaired in patient 1 cells. (A) After radiolabeling of mtDNA-encoded subunits, control or patient 1 fibroblasts were chased for various times as indicated. Mitochondria were isolated, solubilized in Triton X-100 and subjected to BN-PAGE followed by phosphorimage analysis. CI, CIII₂, CIV and CV are indicated. The asterisk (^*) denotes the ~830 kDa intermediate complex. (B) 2D-PAGE analysis of radiolabeled mtDNA-encoded subunits from both control (left panels) and patient 1 (right panels) fibroblasts at different chase times. Mitochondria were isolated, solubilized in Triton X-100 and subjected to BN-PAGE in the first dimension, followed by SDS–PAGE in the second dimension and phosphorimager analysis. Positions of the ~460 and ~830 kDa assembly intermediates and the respiratory chain complexes are shown. The positions of ND1 and ND2 are indicated. CI=complex I, CV=complex V, CIII₂=complex III dimer, CIV=complex IV, CIV_i=complex CIV intermediate, SC=CI/CIII₂ supercomplex.

Radiolabeled mtDNA-encoded translation products from both control and patient 1 fibroblasts were also analyzed after various chase times using two dimensional (2D)-PAGE analysis (Figure 5B). ND2 assembled into the ~460 kDa intermediate complex in control mitochondria but in patient 1 mitochondria, ND2 was absent (Figure 5B, top panels). Before the chase (‘0 h chase'), ND1 was found in the ~830 kDa intermediate complex in control mitochondria but in patient 1 it appeared to stall in an ~400 kDa complex (Figure 5B, top panels). This stalled complex was still present in patient 1 mitochondria at 3 h chase (Figure 5B, middle panels). After 24 h chase, ND1 in control mitochondria had almost fully assembled into complex I or its supercomplex whereas in patient 1 mitochondria, it was completely absent (Figure 5B, bottom panels). SDS–PAGE analysis was also performed, confirming the loss of ND2 between 0 and 1 h chase and the loss of ND1 after 24 h chase in patient 1 mitochondria (see Supplementary Figure 2). These results suggest that the complex I defect in patient 1 cells is due to an impaired assembly of complex I subunits into early intermediates.

Genetic analysis of the CIA30-deficient patient

Genomic DNA sequencing analysis of the patient with CIA30 deficiency revealed two heterozygous mutations in exon 3 of the NDUFAF1 gene, c.1001A>C and c.1140A>G, named according to their cDNA nomenclature (Figure 6A). The c.1001A>C mutation was maternally inherited and predicts a Thr to Pro substitution (T207P) while the c.1140A>G mutation was paternally inherited and predicts a Lys to Arg substitution (K253R). Both Thr 207 and Lys 253 are highly conserved across different species (Figure 6B). The c.1001A>C and c.1140A>G mutations are not known SNPs and were absent from 106 control alleles sequenced (data not shown).

Open in a separate window

Figure 6

Genetic screening of NDUFAF1 in patient 1. (A) Chromatographic trace of relevant NDUFAF1 sequence from control, patient 1 and the patient's parents. The patient is compound heterozygous for a c.1001A>C and a c.1140A>G mutation, the patient's mother was heterozygous for the c.1001A>C mutation while the patient's father was heterozygous for the c.1140A>G mutation. (B) Amino-acid sequence alignment of the mutated region of CIA30 from Homo sapiens (human, accession #Q9Y375), Bos taurus (bovine, accession #XP_615202), Mus musculus (mouse, accession #NP_081451), Xenopus tropicalis (xenopus, accession #NP_001017153.1), Drosophila melanogaster (accession #NP_651718.1), Caenorhabditis elegans (accession #NP_506361.1) and N. crassa (accession #O42636). Shading represents identical residues in four or more species. (C) Predicted NDUFAF1 gene structure showing the location of the mutations within exon 3 (top) and the predicted donor splice sites in control and patient 1 exon 3 (bottom). Predicted percentage scores for splice site probability are shown. SpliceSiteFinder predicts that the c.1140A>G mutation moves the preferred donor splice site 6 bp upstream in patient 1 resulting in the deletion of 6 bp of coding sequence. (D) PCR analysis using primers located in NDUFAF1 exons 2 and 4 of cDNA from control and patient 1 fibroblasts grown without and with cycloheximide (CHX). Patient cDNA shows a second smaller fragment in addition to the expected 333 bp fragment, consistent with the splicing defect predicted by SpliceSiteFinder. (E) Diagrammatic representation of the three sequences identified by cloning the NDUFAF1 cDNA fragment spanning exons 2–4 from patient 1 cDNA. Of 21 sequenced clones, five cDNA sequences represented the expected maternal allele carrying the c.1001A>C (Thr to Pro) mutation. Of 16 paternally derived cDNA sequences (containing c.1001A), four contained the c.1140A>G (Lys to Arg) mutation while 12 lacked the final 6 bp encoded by exon 3 (loss of Val and Lys), consistent with the predicted splicing defect.

cDNA sequencing confirmed the maternally inherited heterozygous mutation at c.1001 and the presence of double sequence suggested a deletion/insertion event had occurred 6 bp upstream of the exon 3/4 boundary (not shown). The c.1140A>G mutation is positioned in the splice site consensus sequence at the end of exon 3 and was suspected to be causing a splicing defect. The NDUFAF1 exon 3 and intron 3 sequence was analyzed using the SpliceSiteFinder prediction program (http://violin.genet.sickkids.on.ca/~ali/splicesitefinder.html), which predicts the presence and relative efficiencies of donor and acceptor sites and assigns them a probability value from 0 to 100 (Shapiro and Senapathy, 1987). The c.1140A>G mutation is predicted to give a preferred donor splice site 6 bp upstream of the wild-type site (Figure 6C). This alternative splice site would lead to a 6 bp deletion of the cDNA, predicting the loss of Val252 and Lys253, which appear evolutionarily conserved in higher eukaryotes (Figure 6B).

cDNA was prepared from control and patient cells and PCR-amplified using primers located in exons 2 and 4. Control cDNA showed a single amplicon of the predicted size but patient cDNA also showed an additional smaller amplicon, consistent with the predicted small deletion being present (Figure 6D). The smaller amplicon was present in cells grown with and without cycloheximide, suggesting the apparent splicing defect did not cause nonsense-mediated mRNA decay (Amrani et al, 2006). To confirm that these results reflected the predicted splicing defect, 21 individual cDNAs were cloned and sequenced. Three different cDNA sequences were identified and these are depicted schematically in Figure 6E. Five cDNA sequences represented the expected maternal allele and contained the c.1001A>C mutation. Two different paternally derived cDNA sequences (containing c.1001A) were identified; four of these contained the c.1140A>G mutation whereas 12 lacked the final 6 bp encoded by exon 3, consistent with the predicted 6 bp deletion encompassing c.1136 to c.1141 (Figure 6E).

CIA30 and mitochondrial disease

The identification of N. crassa CIA30 lead to the screening for genetic defects in the orthologous human NDUFAF1 gene in 13 complex I-deficient patients without identified mutations in genes encoding complex I subunits; however, none were found (Janssen et al, 2002). By screening patient cell extracts with CIA30 antibodies using Western blot analysis, we were able to detect a patient with significantly reduced levels of CIA30 protein. Subsequent sequence analysis of NDUFAF1 transcripts in patient cells indicate that a c.1140A>G mutation leads to mis-splicing, resulting in a 6 bp deletion in ~75% of paternally derived transcripts (deletion of V252 and K253). The remaining ~25% of paternal transcripts encode CIA30 with a K253R mutation. In contrast, the maternally derived c.1001A>C mutation results in a T207P change in CIA30. As patient 1 fibroblasts are deficient in CIA30, the T207P and deletion mutations are likely to cause increased turnover of the protein. The K253R variant may also be unstable but given the relatively mild substitution of one positively charged amino acid for another, it is plausible that this variant may account for the residual protein and complex I assembly seen in patient cells. Introduction of normal CIA30 into patient cells through a lentiviral-mediated approach restored complex I to levels similar to control. In contrast, overexpression of CIA30 in control cells did not lead to increased steady-state complex I levels (data not shown). Thus, our results indicate that the mutations seen in NDUFAF1 are responsible for the defect in complex I and the resulting disease pathology. Our results are consistent with those of Vogel et al (2005) who found complex I levels can be reduced following knockdown of NDUFAF1 using RNA interference.

Most patients with nuclear gene mutations impacting on complex I present in infancy or early childhood and have severe disease often leading to death in early childhood (Kirby et al, 1999). At the time of this report, patient 1 is alive and medically stable at 20 years of age although he displays myopathy, visual impairment, kyphoscoliosis and mild intellectual handicap (a full patient history is provided as Supplementary material). Presumably the presence of some CIA30 is sufficient to promote the assembly of residual complex I seen in the patient cells thereby leading to the ‘relatively' mild phenotype displayed. In addition, since we only assayed CIA30 levels in fibroblasts and lymphoblasts, we cannot exclude that other cells have different splicing and/or protease activities that result in different CIA30 levels in other tissues.

Cloning

The cDNA encoding CIA30 (accession number NM_016013) was obtained from the IMAGE Consortium. For bacterial expression of CIA30, the sequence encoding the predicted mature protein (residues 25–327) was cloned into pQE30 (Qiagen) at BamHI and SmaI sites. For construction of pF5UAS-CIA30, the entire CIA30 open reading frame was cloned into pF5UAS at BamHI and XbaI sites.

Cell culture

HEK-293T cells were cultured in DMEM (Invitrogen) supplemented with 5% (v/v) fetal calf serum (FCS) and penicillin/streptomycin. Patient and control primary fibroblasts and Epstein–Barr virus-transformed lymphoblasts were cultured in DMEM, containing 10% (v/v) FCS supplemented with penicillin/streptomycin and 50 μg/ml uridine. All cells were incubated at 37°C under an atmosphere of 5% CO₂ and 95% air.

In vitro protein import into isolated mitochondria

Plasmid DNA or PCR products were used for in vitro transcription using SP6 RNA polymerase (Promega). In vitro translation of the RNA transcripts was performed according to Ryan et al (2001). CIA30 and pOTC (Peralta et al, 1993) were translated in rabbit reticulocyte lysate in the presence of ³⁵S-methionine/³⁵S-cysteine (EXPRE³⁵S³⁵S Protein Labeling Mix; Perkin Elmer Life Sciences, USA) and incubated with isolated mitochondria at 37°C for various times as indicated. Proteinase K treatment and dissipation of Δψ_m were performed according to Ryan et al (2001).

Blue-native PAGE

BN gel electrophoresis was performed as described previously (Schagger and von Jagow, 1991) with minor modifications (McKenzie et al, 2007). Mitochondrial pellets (50–100 μg protein) were solubilized for 30 min on ice in 50 μl of 20 mM bis–Tris pH 7.4, 50 mM NaCl, 10% (v/v) glycerol containing either 1% (w/v) digitonin (Merck), 1% (w/v) Triton X-100 (Sigma) or 1% (w/v) DDM (or lauryl maltoside; Sigma). After removal of insoluble material by centrifugation, BN-PAGE loading dye (final concentrations: 0.5% (w/v) Coomassie Blue G, 50 mM -amino n-caproic acid (Sigma), 10 mM bis–Tris pH 7.0) was added and protein complexes separated on a 4–13% acrylamide–bisacrylamide BN-PAGE gel. Thyroglobulin (669 kDa), ferritin (440 kDa) and bovine serum albumin (140 and 67 kDa) were used as markers.

Antibodies

Monoclonal antibodies against the complex I subunits NDUFS3 (30 kDa), NDUFS7 (20 kDa/PSST), NDUFB6 (17 kDa), NDUFS5 (15 kDa) and NDUFA6 (14 kDa), the complex III subunit core 1 and the complex II subunit 70 kDa were from Molecular Probes. Antibodies against CIA30, Tom40, NDUFA9 and NDUFB6 were raised in rabbits as described previously (Johnston et al, 2002). The ND1 antibody was a gift from A Lombes (Paris, France) and the mtHsp70 antibody from Nick Hoogenraad (Melbourne, Australia). For detection of complex I on BN-PAGE, antibodies against NDUFA9 were used.

Radiolabeling of mtDNA-encoded translation products

Mitochondrial translation products were labeled as described previously (Chomyn, 1996; McKenzie et al, 2007). The individual mtDNA-encoded subunits were identified based on their size estimates, previous assignments (Chomyn, 1996), 2D-PAGE analysis and their migration with endogenous forms seen with Western blot analysis.

Construction of lentiviral vectors and infection

Inducible lentiviral vectors were based on the pF lentiviral vectors (Naldini et al, 1996) that were a kind gift of I Verma (San Diego, USA). pFUW was modified by introducing a PGK hygromycin-resistance cassette generated by PCR. The ERT2 tamoxifen regulatable ER (Casanova et al, 2002) was a kind gift of T Mantamadiotis (Melbourne, Australia) and was cloned as a fusion with Gal4 VP16 into pFU-PGK-HygroW to generate pFUGEV16. pFUW was also modified by introducing an SV40 puromycin-resistance cassette. This vector was further modified by replacing the ubiquitin promoter with a 5 × multimerized Gal4-binding site to generate pF5UAS. Full sequence details can be obtained upon request.

HEK-293T cells were grown on 10 cm plates to a confluency of 60% and transfected using LipoFectamine2000 (Invitrogen) according to the manufacturer's instructions. One plate was transfected with 5 μg pCMV-δR8.2 (structural gene expression), 2 μg pCAG4-E20 and 3 μg of pF5UAS-CIA30 while the other plate was transfected with 5 μg pCMV δR8.2, 2 μg pCAG4-E20 and 3 μg of pFUGEV16 (expression vector containing transcriptional activator). The supernatants containing packaged virus were harvested 48 h after transfection, combined and filtered through a 0.45 μM membrane filter. Hexabromide (Sigma) was then added to a final concentration of 5 μg/ml and the supernatant was overlaid onto patient fibroblasts for infection. At 72 h later, double infected cells were selected for by the addition of 1 μg/ml puromycin (Sigma) and 0.1 mg/ml hygromycin B (Sigma). Cells were grown in selection media for a period of 72–96 h with media replenished every 24 h. CIA30 expression was induced by addition of 100 nM 4HT to the media. Media containing 4HT was replenished every 24 h.

Genetic analysis

The five exons of NDUFAF1 were amplified from genomic DNA using primers that incorporated at least 30 bp of 5′ and 3′ intronic sequence. In addition, the entire coding region of NDUFAF1 was amplified from cDNA. All amplicons were sequenced using BigDye Terminator Cycle Sequencing kit (Applied Biosystems). For cloning of patient 1 partial cDNA spanning the exon 3/4 boundary, a 333 bp PCR fragment was generated using a forward primer in exon 2 (5′ CGAAGTGGGTACTGTGCAAT 3′) and a reverse primer in exon 4 (5′ TTTATCAGCCAAGGTGAATCC 3′.

We are grateful to the family for participation in this study. We thank J Hoogenraad for technical expertise, A Ohtake for helpful suggestions and A Lombes, T Mantamadiotis, I Verma for reagents This work was supported by grants (MTR, DRT, JS), Postdoctoral Fellowships (MMcK, DMK), RD Wright Fellowship (JS) and a Principal Research Fellowship (DRT) from the Australian National Health and Medical Research Council and a grant from the Australian Research Council (MTR). CJRD and ML are supported by Australian Postgraduate Research Scholarships and CS by a University of Melbourne Postgraduate Research Scholarship.