Author + information
- Received November 21, 2002
- Accepted January 22, 2003
- Published online May 21, 2003.
- Robert W Taylor, PhD*,
- Carla Giordano, MD†,
- Mercy M Davidson, PhD†,
- Giulia d’Amati, MD, PhD‡,
- Hugh Bain, MD§,
- Christine M Hayes, BSc*,
- Helen Leonard, MD§,
- Martin J Barron, PhD*,
- Carlo Casali, MD, PhD∥,
- Filippo M Santorelli, MD¶,
- Michio Hirano, MD†,
- Robert N Lightowlers, PhD*,
- Salvatore DiMauro, MD† and
- Douglass M Turnbull, MD, PhD*,* ()
- ↵*Reprint requests and correspondence:
Prof. Douglass M. Turnbull, Department of Neurology, The Medical School, Framlington Place, University of Newcastle upon Tyne, Newcastle upon Tyne, NE2 4HH, United Kingdom.
Objectives The purpose of this study was to understand the clinical and molecular features of familial hypertrophic cardiomyopathy (HCM) in which a mitochondrial abnormality was strongly suspected.
Background Defects of the mitochondrial genome are responsible for a heterogeneous group of clinical disorders, including cardiomyopathy. The majority of pathogenic mutations are heteroplasmic, with mutated and wild-type mitochondrial deoxyribonucleic acid (mtDNA) coexisting within the same cell. Homoplasmic mutations (present in every copy of the genome within the cell) present a difficult challenge in terms of diagnosis and assigning pathogenicity, as human mtDNA is highly polymorphic.
Methods A detailed clinical, histochemical, biochemical, and molecular genetic analysis was performed on two families with HCM to investigate the underlying mitochondrial defect.
Results Cardiac tissue from an affected child in the presenting family exhibited severe deficiencies of mitochondrial respiratory chain enzymes, whereas histochemical and biochemical studies of the skeletal muscle were normal. Mitochondrial DNA sequencing revealed an A4300G transition in the mitochondrial transfer ribonucleic acid (tRNA)Ilegene, which was shown to be homoplasmic by polymerase chain reaction/restriction fragment length polymorphism analysis in all samples from affected individuals and other maternal relatives. In a second family, previously reported as heteroplasmic for this base substitution, the mutation has subsequently been shown to be homoplasmic. The pathogenic role for this mutation was confirmed by high-resolution Northern blot analysis of heart tissue from both families, revealing very low steady-state levels of the mature mitochondrial tRNAIle.
Conclusions This report documents, for the first time, that a homoplasmic mitochondrial tRNA mutation may cause maternally inherited HCM. It highlights the significant contribution that homoplasmic mitochondrial tRNA substitutions may play in the development of cardiac disease. A restriction of the biochemical defect to the affected tissue has important implications for the screening of patients with cardiomyopathy for mitochondrial disease.
Over the last decade, mutations in the mitochondrial deoxyribonucleic acid (mtDNA) have become increasingly recognized as important causes of disease (1,2), with the number of different pathogenic defects now described exceeding 200 (3). Many take the form of mtDNA rearrangements, but an increasing number of maternally inherited mtDNA point mutations have been described, affecting protein coding, ribosomal ribonucleic acid (mt-rRNA), or more commonly, transfer ribonucleic acid (mt-tRNA) genes (4). Recent epidemiologic studies have estimated the incidence of mtDNA mutations to be at least 1 in 8,000 in a stable Caucasian population within the Northeast of England (5). The majority of reported pathogenic mtDNA mutations are heteroplasmic, a situation in which both the mutated and wild-type forms of mtDNA are present within the same cell. Homoplasmic mtDNA mutations (in which all copies of the genome are affected) have been reported in association with tissue-specific disease, such as Leber’s hereditary optic neuropathy (LHON) and sensorineural hearing loss, affecting protein-encoding (6,7), 12S mt-rRNA (8), and the mt-tRNASer(UCN)(9–11)genes.
A number of mtDNA point mutations have previously been described in patients presenting with cardiomyopathy (12,13). In this report, we describe two families with hypertrophic cardiomyopathy (HCM) associated with a homoplasmic A4300G mutation in the mt-tRNAIlegene. In both families, the clinical features and associated histochemical and biochemical abnormalities were confined to the heart. We were able to study the cardiac tissue of affected individuals from both families, which revealed not only severely decreased respiratory chain activity but also very low steady-state levels of mt-tRNAIle.
The family tree is reported in Figure 1A. Patient no. IV-02 was admitted at age 14 months following a five-day history of a sore throat and swollen glands and a one-day history of lethargy. His development up to this stage was normal. He was in cardiac failure with poor peripheral perfusion. An echocardiogram showed gross dilation of the left ventricle (LV), with grossly diminished global function (Table 1). The LV was hypertrophied. He continued to deteriorate and died of a cardiac arrest two days after admission. Pathologic examination revealed marked concentric hypertrophy of the LV, with no evidence of myocarditis but with hypertrophied myocytes.
The proband (Patient no. IV-03) was carefully monitored from birth. Left ventricular hypertrophy was noted at 11 months, but LV function was good. He was followed until the age of five years at regular intervals; clinical examination, electrocardiography, and echocardiography confirmed biventricular hypertrophy, but with preserved function. The child himself was not restricted in his activities. At five years, he was admitted as an emergency with a four-day history of lethargy. He was in cardiac failure with very poor ventricular function (Table 1). His condition rapidly worsened. He died eight days after admission, despite full supportive therapy. Both surviving boys (Patient nos. IV-01 and IV-04) are asymptomatic. They have evidence of mild LV hypertrophy (both IV-01 and IV-04) and dilation (IV-01) based on both clinical and echocardiographic measures, although neither boy has experienced episodes of cardiac failure. Their mother and maternal grandmother are asymptomatic, and the mother has a normal electrocardiogram (ECG) and echocardiogram. There is no other history of cardiomyopathy in the family.
The detailed clinical features of this large family have been reported previously (14,15). Briefly, a diagnosis of nonobstructive idiopathic HCM was made at 20 years in the proband (IV-01), who underwent heart transplantation 14 years later for congestive heart failure. A maternal cousin (IV-09) with LV hypertrophy progressed to congestive heart failure in six months (Table 1) and is now waiting for heart transplantation with ventricular mechanical assistance. Family history was highly suggestive of a maternal pattern of transmission of the disease (Fig. 1B). All affected members, both symptomatic and asymptomatic, had echocardiographic and ECG features of nonobstructive HCM (Tables 1 and 2). ⇓Neurological examination was normal.
Fibroblast cultures were obtained by explant culture of skin biopsy samples from the probands of both families. Myoblast cultures from the proband of family 1 (IV-03) were obtained as described elsewhere (16). In the absence of primary skeletal muscle cultures, fibroblasts from the proband of family 2 (IV-01) and a cousin (IV-03) were converted to myoblasts using the muscle differentiation gene, MyoD, as described by Sancho et al. (17). Fibroblasts were grown in minimal essential medium (MEM) (Life Technologies, Paisley, United Kingdom) supplemented with 10% fetal calf serum (FCS), 1% MEM nonessential amino acids, and 1% MEM vitamins. Myoblasts were grown in Hams F10 (Life Technologies) supplemented with 20% FCS and 1% chick-embryo extract (ICN Flow Biomedicals, Irvine, California). In addition, both growth media contained 100 μg/ml streptomycin and 100 U/ml benzylpenicillin and were further supplemented with sodium pyruvate (110 μg/ml) and uridine (50 μg/ml) to maintain respiratory-deficient cells.
Histochemical and biochemical analyses of respiratory chain enzymes
Heart tissue from the left and right ventricles of the probands from families 1 and 2 was obtained at autopsy (within 2 h of death) and heart transplantation, respectively. The samples were frozen in isopentane previously cooled to −190°C in liquid nitrogen. Cryostat sections (8 μm) were assayed for succinate dehydrogenase (SDH) and cytochrome coxidase (COX) activities (18). In some cases, sections were dual-stained for both activities.
Respiratory chain enzyme activities were evaluated in mitochondrial fractions (19)prepared from fresh skeletal muscle, cardiac samples, and cultured skin fibroblasts from the proband of family 1 and whole-cell lysates of cultured skin fibroblasts from the proband of family 2. The activities of the individual respiratory chain complexes were determined spectrophotometrically, as previously described (19), using the matrix marker citrate synthase as a standard.
Molecular genetic analyses
Total DNA was isolated from skeletal muscle, heart (LV), and cultured myoblasts from the proband and from circulating blood lymphocytes obtained from his mother, maternal grandmother, and a maternal cousin, according to standard procedures.
Total DNA was isolated from skeletal muscle, heart, circulating lymphocytes, and cultured skin fibroblasts from the proband and from circulating blood lymphocytes from 23 family members spanning four generations of this large pedigree and included affected and nonaffected individuals.
The sequence of the entire mitochondrial genome in the patients’ heart DNA samples was determined using 28 M13-tailed oligonucleotide primer pairs (20). Overlapping polymerase chain reaction (PCR)-amplified fragments were sequenced with BigDye terminator cycle sequencing chemistries on an Applied Biosystems (Foster City, California) 377 automated DNA sequencer. Sequence data were analyzed using Navigator and Factura software (Applied Biosystems) and compared with the revised Cambridge reference sequence (21).
Quantification of mutated mtDNA
The levels of mutated mtDNA in skeletal muscle, heart, cultured cells, and circulating lymphocytes from members of both families were determined by PCR/restriction fragment length polymorphism (RFLP) analysis. Two different strategies were employed. The first, originally described by Casali et al. (14), was used to determine the amount of mutated mtDNA in members of family 2. In the second, a 239–base pair (bp) fragment encompassing the mutation site was amplified with the forward primer L4084 (positions 4084–4103) 5′-GTCACCAAGACCCTACTTCT-3′ and the reverse mismatch primer H4322 (positions 4322–4302) 5′-GGGGGTTTAAGCTCCTATGAT-3′ (mismatch base shown in bold). Samples were subjected to 30 cycles of amplification with an annealing temperature of 56°C; the final extension proceeded for 8 min. After the addition of 30 pmol of each primer, 5 μCi (alpha-32P)-deoxycytidinetriphosphate (3,000 Ci/mmol), and 1 U Taqpolymerase, the PCR reactions were subjected to an additional cycle of amplification. Labeled products were precipitated and equal amounts (1,000 to 2,000 counts) digested with 10 U HinfI (Roche Molecular Biochemicals, Lewes, United Kingdom). Restriction fragments were separated by 12% nondenaturing polyacrylamide gel electrophoresis, dried onto a support, and analyzed with ImageQuant software (Molecular Dynamics, Eugene, Oregon) following exposure to a PhosphorImager (Molecular Dynamics). The mismatch primer creates an additional HinfI recognition site in the PCR product in the presence of the mutation. On digestion, a single HinfI site in the wild-type product generates fragments of 180 and 59 bp. The A4300G mutation generates an additional recognition site cleaving the 180-bp fragment into two smaller products of 158 and 22 bp.
High-resolution Northern blot analysis
Total cytosolic RNA was isolated from solid tissues (∼200 mg; frozen in liquid nitrogen and ground to a fine powder), cultured myoblasts, and fibroblasts (1 to 2 × 106cells) using Trizol reagent (Life Technologies). Large RNA species were precipitated by the addition of 10 mol/l LiCl, allowing smaller RNAs (5S tRNA) to be precipitated from the resulting supernatant. Small RNAs (1 μg) were denatured (90°C for 5 min) and separated through a 13%, 8 mol/l urea denaturing polyacrylamide gel using 1× Tris Borate EDTA (TBE) as a running buffer. Separated samples were electroblotted onto GeneScreen-plus membranes (NEN Dupont, Steverage, United Kingdom) in 0.25× TBE and immobilized by ultraviolet cross-linking. Regions of mtDNA encompassing the tRNAIleand tRNALeu(UUR)genes amplified by PCR were used as probes for Northern blots. The mt-tRNAIleprobe was amplified using the forward primer L4152 (positions 4152–4171) 5′-CGACCAACTCATACACCTCC-3′ and the reverse primer H4328 (positions 4328–4308) 5′-AAATAAGGGGGTTTAAGCTCC-3′. The mt-tRNALeu(UUR)probe was amplified using the forward primer L3200 (positions 3200–3219) 5′-TATACCCACACCCACCCAAG-3′ and the reverse primer H3353 (positions 3353–3334) 5′-GCGATTAGAATGGGTACAAT-3′. Purified PCR products (QIAquick PCR purification columns, Qiagen, Crawley, United Kingdom) were radiolabeled with (alpha-32P)-deoxycytidinetriphosphate (3,000 Ci/mmol) by the random-primer method, and unincorporated nucleotides were removed by gel filtration through a Sephadex G-50 DNA grade column (Amersham Pharmacia Biotech, Uppsala, Sweden). Hybridization was carried out at 42°C overnight in a solution of 5× SSPE, 50% formamide, 10% dextran sulfate, 5× Denhardt’s solution, and 1% sodium dodecyl sulfate (SDS) containing 2 × 106cpm radiolabeled probe. After hybridization, two 15-min washes were performed at room temperature with 2× SSPE, followed by a 15-min wash at 65°C with 2× SSPE and 2% SDS. Blots were subjected to PhosphorImager analysis and the radioactive signal for the mt-tRNAIleprobe (69 bp) normalized to that of the mt-tRNALeu(UUR)probe (75 bp) for each sample.
Histochemical and biochemical analyses of respiratory chain enzymes
Histologic examination of postmortem and explanted hearts failed to show significant myofiber disarray. Histochemical analysis of heart tissue from both probands revealed large numbers of cells devoid of COX activity, which was particularly evident when the sections were dual-stained for both COX and SDH activity. The number of COX-negative cells was higher in the left than in the right ventricle (Figs. 2A to D) . However, the activities of both SDH and COX were normal in skeletal muscle from the proband of family 1, and there were no other histochemical features suggestive of mitochondrial disease, such as the subsarcolemmal accumulation of mitochondria (ragged-red fibers) (Fig. 2E).
The results of the biochemical analysis of respiratory chain enzyme activities for Patient no. IV-03 of family 1 are reported in Table 3. There was a severe defect in the activities of both complexes I and IV in heart mitochondria, with normal activity of complex II. The activities of individual respiratory chain complexes were entirely normal in the muscle (Table 3) and cultured skin fibroblasts from this family, as well as in the fibroblasts and MyoD-converted myoblasts from family 2 (data not shown).
Molecular genetic analysis
Sequence analysis of the entire mitochondrial genome
There was no evidence of mtDNA rearrangements or mtDNA depletion on Southern blot analysis of DNA from the heart (data not shown). The sequence of the entire mitochondrial genome amplified using DNA extracted from both proband hearts showed, in addition to the A4300G mutation, a number of neutral polymorphisms (3,22)that were clearly different between the two families. Consequently, these two families are certainly unrelated. Surprisingly, family 2 showed 10 sequence changes (A3358T, C6336T, T7657C, A8440G, G8790A, A13434G, T13500C, A14062G, G14305A, G14323A) that are not registered on the MITOMAP data base. The presence of such a large number of sequence variants in mitochondrial heart DNA prompted us to also sequence mtDNA from skin fibroblasts, where identical changes were seen. This confirmed not only that the A4300G mutation is the only mtDNA defect common to the two families, but also that a second mitochondrial mutation was not contributing to the expression of the cardiac-specific phenotype.
Demonstration of homoplasmic A4300G mutation
We assessed the level of the A4300G mutation in all available DNA samples from both families using PCR-RFLP analysis. The presence of the mutant G4300 allele creates an additional recognition site for the restriction enzyme HinfI, permitting the detection of heteroplasmic mtDNA at this site. The mutation was homoplasmic in every tissue investigated from Patient no. IV-03 of family 1 (Fig. 3), with no wild-type mtDNA visible even after prolonged exposure of the dried gel to the PhosphorImager plate. The 4300G allele was also homoplasmic in blood cells from the patient’s mother, maternal grandmother, and also a maternal cousin (data not shown). Our findings in family 1 prompted us to reinvestigate the A4300G mutation in family 2, previously reported by Casali et al. (14,15)as heteroplasmic. We screened blood and fibroblast DNA from 23 maternal relatives by PCR-RFLP analysis (Fig. 1). The mutation was homoplasmic in every individual, both symptomatic and asymptomatic, contrary to the earlier report (15)(Fig. 4, RFLP patterns of three representative family members shown). This highlights the importance of using appropriate controls in the analysis of mtDNA mutations (23)and raises concerns about reports of mutations within single families.
Determination of steady-state mt-tRNAIlelevels in both families
To investigate the effect of the A4300G mutation on the processing of mt-tRNAIlefrom its precursor, the steady-state level of mt-tRNAIlewas determined in tissues from affected individuals by Northern blot hybridization. There was a marked decrease (5% to 10% of controls) in the quantity of the mature mt-tRNAIletranscripts from LV tissue of both probands compared with controls (Fig. 5). However, the mt-tRNAIlegene did not show any obvious size change in the mutant form. A similar decrease (10% to 15% of controls) was also observed in skeletal muscle, a clinically unaffected tissue, from the proband of family 1, when compared with steady-state levels of mt-tRNALeu(UUR)in the same samples. Cultured skin fibroblasts and myoblasts from both families expressed the molecular defect to a lesser extent (∼40% and 50% of control values).
We have described the detailed clinical, histochemical, biochemical, and molecular genetic investigation of two families with maternally inherited HCM that revealed a homoplasmic mutation in the mt-tRNAIlegene. The mutation affects a base pair in the secondary structure of mt-tRNAIlein a region of the gene that shows high evolutionary conservation. The clinical features of this cardiomyopathy are distinct from those of familial autosomal-dominant HCM. All affected patients from both families presented with nonobstructive HCM without symptoms of other system dysfunction. There was a mildly asymmetric pattern of LV hypertrophy, with some patients showing a prominent thickening of the posterior wall and others of the ventricular septum (Table 1). The illness often had an adverse clinical course, with LV dilation and failure, even at a young age. There were two childhood deaths in family 1 and two deceased members, a heart transplant recipient at age 34 years and a 23-year-old patient with heart failure waiting for heart transplant in family 2. An additional member of family 2 (Patient no. IV-20) with a LV cavity size within the upper normal limits, showed LV dilation and decreased ejection fraction on a six-month follow-up echocardiogram (Table 1). Although rapidly evolving cardiac failure was observed, neither ventricular arrhythmias nor sudden death were noted in these families.
Numerous heteroplasmic point mutations within mt-tRNA genes have been reported as causing disease, several in association with a number of cardiac muscle disorders (12,13). These include pathologic changes at A4317G (24), C4320T (25), A4295G (26), and A4269G (27)within the mt-tRNAIlegene. Although the A4300G mutation was initially judged to be heteroplasmic when first described (14), our present studies of these two families not only confirm the importance of the mt-tRNAIlegene as a “hot spot” for mitochondrial cardiomyopathy mutations, but also highlight the role that homoplasmic mt-tRNA mutations can play in causing isolated cardiomyopathy. This has important implications for the diagnosis of these conditions. For example, homoplasmic mutations in mtDNA previously reported as neutral polymorphisms could be pathogenic, especially if associated with tissue-specific disorders.
The results of our biochemical and molecular studies in these two families strongly support the pathogenic role of the homoplasmic A4300G mtDNA transition. There were very low activities of respiratory chain complexes I and IV (both contain mitochondrially encoded polypeptide subunits) in heart tissue and a marked decrease in the amount of mature cardiac mt-tRNAIle(∼10% of controls). Surprisingly, comparably low steady-state levels of mt-tRNAIlewere also observed in skeletal muscle, even though only the heart was biochemically and clinically affected. The presence of additional local factors, such as ventricular hemodynamic load or expression of cardiac-specific genes, acting independently or synergistically with the homoplasmic mtDNA mutation could explain the difference in phenotypic expression between cardiac and skeletal muscle. The finding of a marked difference in histochemical COX activity, even between the left and right ventricles, reinforces this hypothesis. The expression of several cardiac genes may vary between the left and right ventricles, both in normal conditions and in response to differences in load and local stress (28–30). The energy demands placed on the LV are much greater than those placed on the right ventricle, and the severity of the mitochondrial defect in the LV of these hearts would be sufficient to cause heart failure.
Homoplasmic mtDNA mutations present many challenges in understanding the molecular mechanism of disease, particularly as they predominantly appear to present in a tissue-specific manner—for example, LHON (6)and sensorineural deafness (8,9). The former is a mitochondrial DNA disorder in which the vast majority of patients present solely with optic neuropathy. The other feature in common between patients with LHON mutations and our families is that in both there was a very sudden onset of symptoms. In LHON, this presents with acute or subacute visual loss and in our patients with cardiac failure. Although it seems likely that the cause of the acute symptoms is an energy crisis in specific tissues, dissecting out the important genetic and environmental factors in LHON has proved extremely difficult. In some patients with sensorineural deafness, there is a clear association between a point mutation (A1555G) in the 12S mt-rRNA gene and toxicity to aminoglycosides (8), an obvious environmental factor. Such a situation, however, is not evident in our patients or in the majority of families with LHON.
Our extensive studies of these families have confirmed that the A4300G mutation is causative because of the mitochondrial biochemical defects in the affected tissue. Although rare compared with heteroplasmic changes, pathogenic, homoplasmic mtDNA mutations have been reported previously in association with various tissue-specific presentations, in particular the optic nerve in patients with LHON and sensorineural deafness (8,11). We believe that in individual patients, similar homoplasmic changes are not being pursued as potential causes of disease because either we do not recognize a particular phenotype or the mtDNA change seen was found to be homoplasmic and thus thought unlikely to be pathogenic. On this basis, it is possible that homoplasmic mt-tRNA mutations are greatly under-reported as causes of mitochondrial disorders (31), and they may play a more important role in the development of cardiomyopathy than previously thought.
We thank Dr. Patrick Chinnery for help in obtaining the samples, Ms. Winsome Walker for technical assistance with tissue culture, and Dr. Cosimo Napoletano and Dr. Vito Piazza for providing clinical data on some of the patients from Family 2.
☆ This work was supported by grants provided to Drs. Turnbull and Lightowlers from the Wellcome Trust, U.K., to Dr. Davidson from the American Heart Association, and to Dr. Giordano from Telethon-Italia.
- cytochrome coxidase
- hypertrophic cardiomyopathy
- Leber’s hereditary optic neuropathy
- left ventricle or ventricular
- mitochondrial deoxyribonucleic acid
- mitochondrial transfer ribonucleic acid gene for isoleucine
- polymerase chain reaction
- restriction fragment length polymorphism
- succinate dehydrogenase
- Received November 21, 2002.
- Accepted January 22, 2003.
- American College of Cardiology Foundation
- ↵MITOMAP: a human mitochondrial genome database, 2002. http://www.mitomap.org
- Wallace D.C.,
- Singh G.,
- Lott M.T.,
- et al.
- Jun A.S.,
- Brown M.D.,
- Wallace D.C.
- Casali C.,
- D’Amati G.,
- Bernucci P.,
- et al.
- Taylor R.W.,
- Turnbull D.M.
- ↵Tanaka M, Ino H, Ohno K, et al. Mitochondrial mutation in fatal infantile cardiomyopathy. Lancet 1990;336:1452
- Santorelli F.M.,
- Mak S.C.,
- Vasquez-Acevedo M.,
- et al.