Article Text


Mutations in SURF1 are not specifically associated with Leigh syndrome

Statistics from

Editor—Isolated cytochrome c oxidase (COX) deficiency is one of the most frequent causes of respiratory chain defects in humans1 and results in a variety of clinical manifestations including Leigh syndrome (LS), hepatic failure, and encephalomyopathy.2-4 COX, the terminal complex of the mitochondrial respiratory chain, is composed of 13 subunits, three of them being encoded by mitochondrial DNA genes. Nuclear genes encode the 10 other subunits. However, a much larger number of proteins of nuclear origin are required for the correct assembly and function of COX. More than 30 different genetic complementation groups for COX assembly have been isolated in yeast.5 Somatic cell genetic studies have shown that most cases of LS associated with COX deficiency belong to one complementation group.6 7 The gene defect in this group was mapped to chromosome 9q34 and analysis of a candidate gene (SURF1) showed several mutations, all of which predict a truncated protein.2 8

While mutations in the three mitochondrial COX genes,COX I-III, have been reported in a few patients,9-14 no mutations in any of the nuclear genes encoding the COX protein subunits have been identified so far.15 To date, mutations in three nuclear genes have been identified in patients with isolated COX deficiencies:SURF1 mutations in patients presenting with LS,2 8 SCO2 mutations in patients with fatal hypertrophic cardiomyopathy and encephalopathy,16 17 and aCOX10 mutation in one family with mitochondrial encephalopathy presenting with ataxia, severe muscle weakness, ptosis, pyramidal syndrome, and status epilepticus.18 While all three genes are involved in COX assembly, they appear to be associated with three distinct clinical entities. In particular, SURF1 mutations have so far been reported only in LS.19 Similarly, the few patients with mutations in the nuclear gene encoding the flavoprotein subunit of the succinate dehydrogenase have all presented as LS so far.20 21 Thus, in contrast with mutations in mitochondrial DNA, mutations in nuclear genes encoding proteins involved in respiratory chain assembly or function appear to result in more specific phenotypes. However, various clinical presentations associated with mutations in NDUFV1, a gene encoding one of the mitochondrial complex I subunits, have been described.22 Here, we show that new mutations inSURF1 can produce a clinical phenotype with villous atrophy and hypertrichosis as presenting symptoms, without the typical central nervous system pathology associated with LS.

A girl was born at term to unrelated, healthy parents after a normal pregnancy (birth weight 2280 g, length 49 cm, head circumference 32 cm, −3 SD). At 6 months of age, sudden failure to thrive, vomiting, and major hypertrichosis was noted (fig 1). She was first admitted to hospital at 1 year of age. Laboratory investigations, including plasma amino acids, lysosomal enzymes, urinary mucopolysaccharides, and endocrine evaluation, were normal apart from a low level of IgF1. A duodenal biopsy showed partial villous atrophy and she was put on a gluten free diet. Feeding difficulties and vomiting persisted and she was then referred to our unit at 20 months of age. Physical examination showed generalised hypertrichosis, a coarse face, strabismus, and kyphosis. Weight and length were below the mean (weight 8200 g, <−3 SD, length 73 cm, <−3 SD, head circumference 46.5 cm, <−2 SD). She was also hypotonic, was unable to walk, and had poor language development with only a few words. A cerebral CT scan was normal, as were hepatic and cardiac function. Laboratory investigations showed a high level of lactate (3.8 mmol/l, normal <2.5) and a general hypoaminoacidaemia contrasting with normal levels of alanine and proline. Medication with carnitine, vitamin C, and vitamin E was started. She then developed bilateral optic atrophy and psychomotor retardation (no ambulation at 34 months of age) and a profound asthenia. She died at 3 years of age from apnoea.

Figure 1

The patient. Note the hypertrichosis on her back.

Respiratory chain (RC) enzyme analysis was performed on skeletal muscle homogenate and cultured skin fibroblasts of the patient. Cytochrome c oxidase (CIV, EC, antimycin sensitive decylubiquinol cytochrome c reductase (CIII, EC, malonate sensitive succinate quinone dichlorophenolindophenol (DCPIP) reductase (CII, EC1.3.99.1), NADH quinone reductase (CI, EC, and ATPase (CV, EC3.6.1.34) activities were measured spectrophotometrically on skeletal muscle microbiopsies and cultured skin fibroblasts as previously described.23 Results were expressed both as absolute values and as activity ratios.24 Both skeletal muscle homogenate and skin fibroblasts of the patient showed markedly deficient COX activity whereas the activities of the other complexes were in the normal ranges (table 1). This resulted in highly abnormal ratios of COX to other RC enzyme activities. No increase in the activity of the succinate cytochrome c reductase was observed in either tissue, and normal cytochrome c content relative to cytochrome b was measured spectrophotometrically in patient fibroblasts (not shown). The severe COX deficiency found in both skeletal muscle and fibroblasts from this patient is comparable to that previously reported in patients with LS and SURF1mutations.6 7

Table 1

Enzymological investigation of skeletal muscle homogenates and cultures of skin fibroblasts from patient and controls

An immunoblot analysis of mitochondrially enriched fractions from control and patient fibroblasts was performed to investigate the steady state levels of the structural subunits of the COX complex. Western blot analysis of the COX subunits was performed as previously described.26 Quantification of COX subunits was performed using Sigma Gel software. For comparison, the analysis was also performed on a COX deficient patient with typical LS associated with a known mutation in SURF1 (homozygous 550delG mutation). A severely reduced steady state level of several COX subunits was observed in both fibroblast lines, consistent with deficient assembly of COX (fig 2). COXII, COXIII, VIa, and VIc subunits were the most severely affected, as compared to subunits I, IV, Va, and Vb which were present at 30-40% of control levels.

Figure 2

Immunoblots of mitochondrial proteins isolated from patient and control skin fibroblasts. (A) Cytochrome oxidase subunit composition in control (C), patient (P), and typical LS patient (LS) fibroblasts. The upper blots were developed with specific antibodies against the various COX (COXI-VIc) subunits as indicated. The lower blots were developed with monoclonal antibodies against VDAC (voltage dependent anion channel) or the flavoprotein (Fp) of the succinate dehydrogenase (SDH-Fp). (B) Cytochrome oxidase relative to citrate synthase (COX/CS) in control and patient fibroblasts before (P) and after infection with a SURF1 retroviral vector (P+SURF1). (C) Surf1 protein and COX subunits II and IV in patient fibroblasts before (P) and after transfection with a SURF1 retroviral vector (P+SURF1). The lower blot was developed with a monoclonal antibody against VDAC.

PCR amplification and direct sequencing of the nine exons ofSURF1 was performed on genomic DNA from the patient and his parents. The nine exons of theSURF1 gene were amplified from genomic DNA using the following intronic primers. Oligonucleotide sequences (forward/reverse, 5′-3′) of the SURF1 gene were as follows: exons 1 and 2: GATGCAGATGCTTCCTGCGTC, TCAAAGT GCAGGGCAGACAG, exons 3 and 4: AGGGCTTCT GGCTCCATGTCA, CTCAAGTAAAACAGGCCCTA, exon 5: CAAACCTTGCTCGGCCACTGT, TGCCTC TGCCAGGACAGCCA, exons 6 and 7: CCCCACCT GAAGTAGCACTTT, AAGCTACTTGTTCCGAG ATG, exons 8 and 9: AGTAGGGGGTGGACTTG CGT, TTATCCAGGGACAGGGCTTC. Amplification products were electrophoresed through a 2% low melting point agarose gel and directly sequenced using the PRISMTM Ready Reaction Sequencing Kit (Perkin-Elmer) on an automatic sequencer (Applied Biosystems). TheSURF1 cDNA was amplified by RT-PCR from total RNA isolated from patient fibroblasts using primer pairs flanking the open reading frame25 and directly sequenced after gel purification. For further analysis of the splice site mutation, the amplified RT-PCR products were cloned using the TOPOBLAST cloning kit (Invitrogen) and individual plasmids were isolated and sequenced. Two heterozygous mutations were found in the patient: a G to A transition at nucleotide 553 of the cDNA changing an evolutionarily conserved glycine into a glutamic acid (G180E) in exon 6 and a G to C transversion in the acceptor splice site of intron 6 (G603-1C) (fig 3A, B). The G553A mutation was found to be heterozygous in the mother and was absent in the father, whereas the intron 6 splice mutation was only present in the father. Both mutations were absent from eight controls. The cDNA from patient fibroblasts was amplified by RT-PCR. Sequencing of the RT-PCR product showed that the mutation in the acceptor splice site of intron 6 unmasked a cryptic acceptor in exon 7, resulting in an in frame 6 bp deletion at the start of exon 7. This was confirmed by sequence analysis of cloned RT-PCR products (fig 3C). The deletion is predicted to eliminate amino acids I197 and E198.

Figure 3

Analysis of the SURF1 gene and cDNA in the patient.(A) Heterozygous mutation at G553A (cDNA) inherited from the mother predicting a G160E substitution. (B) Heterozygous mutation inherited from the father in the acceptor splice site of intron 6 (G603-1C) in the genomic DNA. (C) In frame deletion of the first six bases of exon 7 in patient cDNA predicting a deletion of I197 and E198 in the Surf1 protein. The sequence shown is a clone derived from the RT-PCR product.

To establish the causative nature of these SURF1mutations in the COX deficiency, we first tested whether microcell mediated transfer of SURF1carrying chromosome 9 would rescue cell growth on selective medium for respiratory competent cells (RPMI with galactose minus uridine and pyruvate).2 Grown under these selective conditions, patient fibroblasts died after three to five days, while chromosome 9 complemented cells were still alive after more than two weeks. The direct involvement of SURF1 was established by transducing patient fibroblasts with a retroviral vector expressingSURF1 cDNA.25 Expression ofSURF1 normalised the steady state levels of COX subunits (fig 2C) and restored COX activity to control levels relative to citrate synthase (CS), a mitochondrial matrix marker (fig2B).

Based on the identification of the disease causing mutations inSURF1 in the proband, prenatal diagnosis was offered to the family. In agreement with normal COX measurable in a trophoblast biopsy of the patient, we found thatSURF1 in the fetal cells did not harbour either of the mutations. This avoided replicating the enzyme test on amniocytes, which is currently required to confirm biochemical diagnosis of respiratory chain deficiency.

The missense and in frame mutations reported inSURF1 in our patient with COX deficiency differ from those previously reported in this gene.2 8 27 Most of the reported mutations are truncating loss of function mutations associated with undetectable Surf1 protein, a homogenous biochemical phenotype, and a typical LS presentation. Missense mutations in SURF1 in COX deficient LS patients appear to be rare as only two have been reported,28 and an in frame deletion has never been described. The clinical phenotype of the patient reported here was also markedly different from the typical LS in patients withSURF1 mutations. The patient presented with villous atrophy, hypertrichosis, and only mild neurological involvement, with no significant brain lesions as shown by the normal CT scan. Villous atrophy as well as hair and skin anomalies have been observed in patients with a mitochondrial disease.29 30In particular, diarrhoea and villous atrophy have been reported in association with mtDNA rearrangements.31 Hair and skin anomalies as a manifestation of respiratory chain deficiency were found in 10% of children (14/141) with mitochondrial disorders in our original cohort.32 Three of these 14 patients presented with sporadic hypertrichosis. The latter was not attributable to hypothyroidism, gross malnutrition, or drugs, and therefore appears to be one of the numerous unexplained symptoms found in association with respiratory chain disorders. Hypertrichosis was also noticed as an unusual presenting symptom of COX deficiency in two boys with COX deficiency.33 It is possible that the subtle neurological signs initially present in our patient might have developed into the typical lesions of LS in her brain had she lived longer; however, typical LS patients develop the distinctive CNS pathology in the first year of life.

It is tempting to ascribe the distinctive clinical presentation to the specific nature of the mutations in this case. The association of different clinical presentations with different classes of mutations in a given gene is a recurrent theme in numerous genetic diseases. However, the biochemical analysis of fibroblasts in our patient showed that the mutations resulted in a severe COX deficiency similar to that observed in typical LS patients. Further, Surf1 protein was undetectable in our patient and immunoblot analysis showed that reductions in the steady state levels of the COX subunits were similar to those in an LS patient. Although mutations inSURF1 are associated with a specific defect in COX, other anomalies of the respiratory chain have been reported in association with SURF1 mutations. In humans, an increased cytochrome c content has been detected in some patients,28 while in a yeast strain deleted forShy1, the SURF1homologue, an increased electron flow to cytochrome c was indicated by an increased succinate cytochrome c reductase.34 Neither of these additional features was noticed in our patient.

How can we explain the variation in the clinical presentation and/or course of the disease associated with SURF1mutations? Although we cannot rule out environmental influences or the influence of modifying genes, we favour a model in which there is tissue specific variability in the stability (and therefore function) of the Surf1 protein that depends on the nature of the mutations. All of the truncating mutations that have been reported inSURF1 in LS patients disrupt the predicted C-terminal transmembrane domain, a part of the protein that mutational analysis has shown to be essential for protein stability.25 The mutations reported here would cause an amino acid substitution or small deletion in the loop (spanning the two transmembrane domains) that is predicted to reside in the mitochondrial intermembrane space. It is not clear why such mutations would completely destabilise the protein. The G180E mutation does change a charge in an amino acid that has been strictly conserved during evolution,28 and this could disrupt membrane insertion of the protein or its interaction with other proteins. However, the identity of the two deleted amino acid residues has not been evolutionarily conserved.

Although the precise function of Surf1 is unknown, it is clearly involved in some way with the assembly or maintenance of the COX complex and thus must be involved in protein-protein interactions.25 27 The nature of these proteins is unknown, but it is possible that they are tissue specific and that some mutants could be partially functional in some tissues (like brain) but not others (like fibroblasts). The COX complex does assemble and function, albeit inefficiently (10-15% of control levels), in the complete absence of Surf1 protein, and it is therefore likely that even a small amount of Surf1 protein could produce a significant increase in COX activity.

Tissue specific deficiencies in the activity of COX have been reported in the context of three other putative COX assembly factors, suggesting that such effects might be common. A biochemically distinct form of COX deficient LS has been mapped to chromosome 2 in a French Canadian population.35 Although the gene defect has not been identified, the cDNAs of the structural COX subunits are normal and the deficiency appears to result from a failure of complex assembly. COX activities are very low in brain and liver of these patients, but relatively high in muscle and fibroblasts.35 Recently, mutations in SCO2, whose gene product is thought to act as a mitochondrial copper chaperone for COX, have been found to cause a distinct form of early onset fatal cardiomyopathy. COX activity is reduced to nearly undetectable levels in heart and skeletal muscle of these patients, but relatively modest decreases are seen in fibroblasts.16 17 Finally, a mutation inCOX10, which encodes haem A:farnesyltransferase, causes mitochondrial encephalopathy without heart or liver involvement.18 The reasons for the tissue specific differences in COX activity associated with mutations in ubiquitously expressed COX assembly factors remain obscure, but the evidence clearly points to alternate pathways for assembly and different protein-protein interactions.

Three clear predictions emerge from the interpretation of the results presented here. (1) Other patients with COX deficiency, but without LS, will be found to harbour SURF1 mutations, but all will have at least one allele that does not result in a truncated protein. (2) Surf1 protein will be detectable in the brain of these patients. (3) Tissue or cell specific interacting protein partners will be found for Surf1.

Sorting out the complexity of the phenotype-genotype relationship in respiratory chain diseases, even when of nuclear origin, remains a formidable challenge. The identification of the molecular bases of respiratory chain dysfunction in these cases is an important practical goal because it offers confident prenatal diagnosis for these devastating diseases, as shown in this study.


This research was supported in part by the Association Française contre les Myopathies and by grants from the MDA and March of Dimes (EAS). JKvKR is recipient of a grant from INSERM, JWT is a recipient of a grant from the Wellcome Trust. EAS is an MNI Killam Scholar.


View Abstract

Request permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.