OPA1 mutations impair mitochondrial function in both pure and complicated dominant optic atrophy

Sir, Autosomal dominant optic atrophy (DOA) is a common cause of inherited visual failure affecting at least 1 in 35 000 of the general population (Yu-Wai-Man et al., 2010a). Pathogenic OPA1 mutations account for about 60% of all cases, causing bilateral, symmetrical optic atrophy secondary to the highly tissue-specific loss of one cell type - the retinal ganglion cell (Lenaers et al., 2009). Although optic nerve degeneration remains the defining feature of DOA, we recently reported in Brain that up to 20% of patients with OPA1 mutations will also develop additional neuromuscular complications including deafness, ataxia, myopathy, peripheral neuropathy and progressive external ophthalmoplegia (Yu-Wai-Man et al., 2010b).

In two earlier studies, also published in Brain (Amati-Bonneau et al., 2008; Hudson et al., 2008), we described for the first time the intriguing association of these syndromal DOA+ variants with multiple mitochondrial DNA deletions and cytochrome c oxidase (COX)-negative skeletal muscle fibres (Zeviani, 2008). Interestingly, these mitochondrial defects were subsequently identified in OPA1 patients with pure optic nerve involvement, but at levels four times lower compared with the DOA+ group (Yu-Wai-Man et al., 2010b). The involvement of other tissue types in DOA+ could therefore be a direct consequence of the greater accumulation of these secondary mitochondrial DNA abnormalities, the latter potentiating an already compromised mitochondrial oxidative reserve due to the mutant OPA1 protein. To determine whether this was the case, we used in vivo phosphorus magnetic resonance spectroscopy (³¹P-MRS) to specifically measure mitochondrial oxidative function in a subgroup of OPA1 patients from our original reports, correlating our findings with the histochemical and mitochondrial DNA defects identified in skeletal muscle biopsies, and thus extending our previous observations (Yu-Wai-Man et al., 2010b).

For this ³¹P-MRS study, we selected 17 patients harbouring 12 different pathogenic OPA1 mutations [mean age = 47.7 years, standard deviation (SD) = 10.6 years, range = 30.0–65.0 years): nine patients with isolated optic atrophy and eight patients with complex neuromuscular phenotypes (Tables 1 and 2). This patient group was compared with 17 newly identified, age-matched, normal controls with no evidence of ocular or neuromuscular pathologies (mean age = 48.3 years, SD = 8.5 years, range 37.0–63.0 years, P = 0.8457). This study had the relevant institutional ethical approval and informed consent was obtained in accordance with the Declaration of Helsinki.

Table 1.

Clinical features of our OPA1 patient cohort

Patient	Age (years)	Sex	FHx	Onseta (years)	Snellen BCVA		Clinical phenotype
					Left	Right	Optic Atrophy	Deafness	Ataxia	Myopathy	Neuropathy	PEO	Others
1	50	M	+	8	CF	CF	+
2	58	F	+	–	20/30	20/30	+		+				MS-like illness
3	30	M	+	5	CF	CF	+		+	+	+	+
4	38	F	+	1	CF	CF	+	+	+	+	+	+	Migraine
5	43	M	+	5	20/200	20/200	+
6	59	M	+	5	20/200	20/200	+
7	43	M	+	13	CF	CF	+		+	+	+	+
8	65	M	+	5	20/200	20/200	+
9	60	F	+	16	CF	CF	+
10	40	F	+	15	20/60	20/120	+
11	44	F	+	15	20/120	20/200	+		+		+
12	54	M	+	5	CF	CF	+	+
13	43	M		5	20/200	20/200	+	+		+		+
14	54	M	+	–	20/20	20/30	+
15	39	M	+	15	20/200	20/120	+
16	59	M	+	5	CF	CF	+				+		HSP, Migraine
17	31	M	+	11	20/40	20/60	+

^aAge of onset of visual failure.

The clinical and molecular characteristics of the eight patients with DOA+ phenotypes have been detailed previously (Yu-Wai-Man et al., 2010b): Patient 2 (Pedigree UK-6), Patient 3 (Pedigree UK-11), Patient 4 (Pedigree UK-11), Patient 7 (Pedigree UK-12), Patient 11 (Pedigree UK-8), Patient 12 (Pedigree UK-1), Patient 13 (Pedigree UK-13) and Patient 16 (Pedigree UK-3). BCVA = best corrected visual acuity; CF = counting fingers; F = female; FHx = family history; HSP = hereditary spastic paraparesis; M = male; MS = multiple sclerosis; PEO = progressive external ophthalmoplegia.

Table 2.

OPA1 mutations and mitochondrial abnormalities identified in skeletal muscle biopsies

Patient	OPA1 mutations				Skeletal muscle biopsy
	cDNA change	Location	Consequence	Domain	COX-negative fibres (%)	Mitochondrial DNA deletions
1	c.2613+1g>a	Intron 25	Splicing defect		0.0	–
2	c.2613+1g>a	Intron 25	Splicing defect		1.4	+
3	c.1635C>A	Exon 17	p.S545R	Dynamin	0.4	+
4	c.1635C>A	Exon 17	p.S545R	Dynamin	N/A	N/A
5	c.2708_2711delTTAG	Exon 27	p.V903fsX3	GED	0.6	+
6	Exons 1-5b deletion		p.M1fsX208		2.0	+
7	c.1294A>G	Exon 13	p.I432V	GTPase	10.0	+
8	c.2818+5g>a	Intron 27	Splicing defect		3.0	+
9	c.2713C>T	Exon 27	p.R905X	GED	3.0	+
10	c.2713C>T	Exon 27	p.R905X	GED	0.5	+
11	c.32+1g>a	Intron 1	Splicing defect		2.0	+
12	c.1212+3a>t	Intron 12	Splicing defect		0.3	+
13	c.1334G>A	Exon 14	p.R445H	GTPase	3.1	+
14	c.1516+1g>t	Intron 15	Splicing defect		1.8	+
15	c.1516+1g>t	Intron 15	Splicing defect		0.0	+
16	c.876-878delTGT	Exon 9	p.V294fsX667		2.1	+
17	c.876-878delTGT	Exon 9	p.V294fsX667		0.1	+

COX = cytochrome c oxidase; GED = GTPase effector domain; GTP = guanosine triphophate; N/A = not available.

Resting ³¹P-MRS parameters were within the normal range for the entire OPA1 group (Table 3). Similarly, OPA1 patients did not exhibit abnormalities in proton handling at rest (pH), during or following exercise (maximum proton efflux rate). The overall recovery of both phosphocreatine (P = 0.0218) and adenosine diphosphate (ADP) (P = 0.0274) to basal levels, as measured by the half time τ_1/2, was significantly delayed in all OPA1 mutational carriers. Subgroup analysis showed a significant difference for patients with both pure DOA and DOA + phenotypes compared with controls, but there was no significant difference in τ_1/2 phosphocreatine and τ_1/2 ADP between these two disease subgroups (Fig. 1A and B). The frequency of COX-negative fibres in skeletal muscle biopsies (mean = 1.9%, SD = 2.4%, range = 0–10.0%, n = 16) did not correlate with either τ_1/2 phosphocreatine (Spearman rank correlation coefficient = 0.0310, P = 0.9094) or τ_1/2 ADP (Spearman rank correlation coefficient = −0.1636, P = 0.5449) (Fig. 1C and D). The correlation between COX deficiency and both τ_1/2 phosphocreatine and τ_1/2 ADP remained non-significant after exclusion of two outlying data points (Fig. 2). For three DOA+ patients (Patients 3, 7 and 16), citrate synthase and mitochondrial respiratory chain enzyme activities were determined in mitochondrially enriched suspensions obtained from homogenized muscle specimens (Taylor et al., 2004). All measurements were within the normal assay range.

Table 3.

Resting and post-exercise ³¹P-MRS measurements for OPA1 patients and controls

	Patients (n = 17)	Controls (n = 17)	P-value
	Mean (95% CI)	Mean (95% CI)
Age (years)	47.7 (42.2–53.1)	48.3 (43.9–52.7)	0.8457
Resting
PCr (mM)	30.9 (29.8–32.0)	31.9 (30.9–32.8)	0.1539
Pi (mM)	2.90 (2.40–3.39)	3.02 (2.81–3.24)	0.6279
PCr/Pi ratio	11.6 (9.7–13.5)	10.7 (9.9–11.6)	0.3803
ADP (µM)	9.79 (9.48–10.09)	9.82 (9.39–10.25)	0.9029
pH	7.05 (7.04–7.06)	7.05 (7.03–7.07)	0.7598
Post-exercise
Initial PCr resynthesis rate (mM/min)	11.8 (7.9–15.6)	10.5 (8.5–12.5)	0.5366
τ1/2 PCr (s)	39.4 (28.0–50.8)	25.9 (22.7–29.1)	0.0218*
τ1/2 ADP (s)	30.1 (21.3–38.8)	20.2 (17.9–22.4)	0.0274*
Maximum proton efflux rate (mmol/l/min)	2.75 (2.09–3.42)	2.55 (2.01–3.09)	0.6157

Phosphorus spectra were acquired from the gastrocnemius and soleus muscles using a 3-T Intera Achieva scanner (Philips, Best, The Netherlands) during: (i) a 1-minute period of rest; (ii) a 3-minute period of plantar flexion at 25% of the maximum voluntary contraction; and (iii) a 6-minute recovery period (Trenell et al., 2006; Hollingsworth et al., 2008). Spectral quantification was performed with the Java-based magnetic resonance user interface (jMRUI v.3.0), using AMARES with custom prior knowledge appropriate to skeletal muscle (Naressi et al., 2001). ADP = adenosine diphosphate; PCr = phosphocreatine; Pi = inorganic phosphate. *Significant P-value.

Figure 1.

Subgroup comparison of post-exercise ³¹P-MRS parameters between patients with pure DOA (n = 9), DOA+ (n = 8) and age-matched normal controls (n = 17), for (A) τ_1/2 phosphocreatine (PCr): *P = 0.0434, ** P = 0.0014, non-significant (NS) at P = 0.6406, and (B) τ_1/2 adenosine diphosphate (ADP): *P = 0.0256, **P = 0.0079, NS at P = 0.5627. Correlation of in vivo markers of metabolic recovery with the frequency of COX-negative fibres identified in skeletal muscle biopsies: (C) τ_1/2 phosphocreatine: Spearman rank correlation coefficient = 0.0310, P = 0.9094, and (D) τ_1/2 ADP: Spearman rank correlation coefficient = −0.1636, P = 0.5449. The error bars represent the standard error of the mean.

Figure 2.

Correlation of in vivo markers of metabolic recovery with the frequency of COX-negative fibres identified in skeletal muscle biopsies, after exclusion of two outlying data points: (A) τ_1/2 phosphocreatine (PCr): Spearman rank correlation coefficient = 0.1718, P = 0.5570, and (B) τ_1/2 ADP: Spearman rank correlation coefficient = −0.1079, P = 0.7134.

OPA1 mutations exert a deleterious effect on in vivo mitochondrial function, irrespective of mutational subtypes and disease severity. Our a priori hypothesis was that the additional neuromuscular manifestations seen in DOA+ patients were the result of a more pronounced biochemical defect. What are the reasons therefore for the lack of a significant difference in bioenergetic impairment between patients with pure DOA and DOA+ phenotypes? Our results are statistically robust and consistent for both phosphocreatine and ADP recovery parameters. It is therefore unlikely that the inclusion of additional patients would influence the outcome of this ³¹P-MRS study. The level of COX deficiency was determined in quadriceps or tibialis anterior biopsies, whereas phosphorus spectral data were acquired from the gastrocnemius and soleus muscles. It is possible that structural and functional differences exist between these various muscle groups, which could have influenced our measurement parameters and comparisons. Notwithstanding this caveat, our observations overall indicate that impaired oxidative phosphorylation is only part of the problem, and different pathways must be mediating retinal ganglion cell loss and cellular dysfunction in other organ systems. It is also intriguing that the same OPA1 mutation can lead to both isolated optic nerve involvement and a more severe form of the disease, with the development of neuromuscular complications (Yu-Wai-Man et al., 2010b). Additional factors are clearly modulating the pathogenic expression of the OPA1 mutation resulting in markedly variable clinical phenotypes. Future studies are required to determine whether DOA+ is related to other pathological consequences triggered by the greater accumulation of these mitochondrial DNA deletions; such as the induction of mitochondrial proliferation, which is known to have a pro-apoptotic effect (Aure et al., 2006; Yu-Wai-Man et al., 2010c), or other unrelated mechanisms linked with mitochondrial network instability, OPA1 being a critical pro-fusion protein (Lenaers et al., 2009).

Our ³¹P-MRS study supports an earlier report of impaired in vivo mitochondrial function in six affected individuals from two families, segregating isolated optic neuropathy and the c.2708_2711del(TTAG) deletion (p.V903fsX3) (Lodi et al., 2004). We have extended this observation to 11 additional pathogenic OPA1 mutations, including for the first time eight patients with DOA+ phenotypes. Previous ³¹P-MRS studies have also confirmed a respiratory chain complex defect in Leber hereditary optic neuropathy (LHON), the classic paradigm of a primary mitochondrial optic neuropathy (Yu-Wai-Man et al., 2009). There seemed to be a mutational hierarchy; the m.11778G>A LHON mutation exhibiting the most pronounced effect on mitochondrial ATP synthesis, followed by m.14484T>C and m.3460G>A (Lodi et al., 1997, 2002). The latter only led to a subtle metabolic deficit, highlighting the vulnerability of retinal ganglion cells to even mild energetic imbalances. Given the pathological similarity shared by LHON and DOA, it is likely that the bioenergetic defect revealed by ³¹P-MRS is contributing to selective retinal ganglion cell loss and visual failure in both disorders.

In three DOA+ patients, biochemical studies performed on skeletal muscle mitochondrial fractions were normal, despite the fact that both their τ_1/2 phosphocreatine and τ_1/2 ADP measurements exceeded the upper 95% control values. These findings suggest that ³¹P-MRS could be a useful diagnostic adjunct in patients with suspected mitochondrial disease, when histochemistry and in vitro biochemical studies are inconclusive. Furthermore, mitochondrial oxidative physiology can be inferred both at rest and crucially following a period of exercise, the post-recovery kinetics being particularly sensitive in detecting a tissue’s capacity to respond to energy demands (Barbiroli et al., 1998).

There is a complex, and still poorly defined, interplay between the multiple cellular functions regulated by the OPA1 protein. However, irrespective of the combination of factors involved, the ultimate loss of mitochondrial membrane potential and the release of cytochrome c molecules are likely to be key final common events triggering apoptotic cell death. Unravelling these intricately linked mechanisms will hopefully contribute to the long-term goal of developing therapeutic interventions, not only for DOA, but also for other mitochondrial optic neuropathies.

Funding

Medical Research Council (MRC, UK) Clinical Research Fellowship in Neuro-ophthalmology (to P.Y.W.M.); Wellcome Trust Senior Fellowship in Clinical Science (to P.F.C.); P.F.C. also receives funding from Parkinson's UK, the Association Française contre les Myopathies, the MRC Translational Muscle Centre, and the UK National Institute for Health Research (NIHR) Biomedical Research Centre in Ageing and Age Related Diseases.