Complex I mutations synergize to worsen the phenotypic expression of Leber ’ s hereditary optic neuropathy

caused by the m.11778G . A mutation. The significant correlation between severe biochemical phenotypes in these mutants cybrids and higher penetrance of LHON in these pedigrees harboring both m.11778G . A and m.3866T . C mutation demonstrated that the m.3866T . C mutation enhanced the penetrance and occurrence of optic neuropathy in the Chinese families carrying both m.3866T . C and m.11778G . Therefore, our

Leber's hereditary optic neuropathy (LHON) is a maternal inheritance of eye disease because of the mitochondrial DNA (mtDNA) mutations. We previously discovered a 3866T>C mutation within the gene for the ND1 subunit of complex I as possibly amplifying disease progression for patients bearing the disease-causing 11778G>A mutation within the gene for the ND4 subunit of complex I. However, whether and how the ND1 mutation exacerbates the ND4 mutation were unknown. In this report, we showed that four Chinese families bearing both m.3866T>C and m.11778G>A mutations exhibited higher penetrances of LHON than 6 Chinese pedigrees carrying only the m.3866T>C mutation or families harboring only the m.11778G>A mutation. The protein structure analysis revealed that the m.3866T>C (I187T) and m.11778G>A (R340H) mutations destabilized the specific interactions with other residues of ND1 and ND4, thereby altering the structure and function of complex I. Cellular data obtained using cybrids, constructed by transferring mitochondria from the Chinese families into mtDNA-less (r o ) cells, demonstrated that the mutations perturbed the stability, assembly, and activity of complex I, leading to changes in mitochondrial ATP levels and membrane potential and increasing the production of reactive oxygen species. These mitochondrial dysfunctions promoted the apoptotic sensitivity of cells and decreased mitophagy. Cybrids bearing only the m.3866T>C mutation displayed mild mitochondrial dysfunctions, whereas those harboring both m.3866T>C and m.11778G>A mutations exhibited greater mitochondrial dysfunctions. These suggested that the m.3866T>C mutation acted in synergy with the m.11778G>A mutation, aggravating mitochondrial dysfunctions and contributing to higher penetrance of LHON in these families carrying both mtDNA mutations.
Leber's hereditary optic neuropathy (LHON) is the most common maternally transmitted eye disorder, characterized by the degeneration of retinal ganglion cells (RGC) and loss of central vision (1)(2)(3)(4)(5)(6). A number of mitochondrial DNA (mtDNA) mutations have been linked to the pathogenesis of LHON with various degrees of contribution (7)(8)(9)(10). These mtDNA mutations included the primary mutations, which can each cause LHON, and the secondary mutations, which may interact with the primary mutation to increase the probability of clinical expression (1,8). In the majority of cases worldwide, LHON was caused by three primary mtDNA point mutations, ND4 11778G.A, ND6 14484T.C, and ND1 3460G.A, affecting subunits of NADH:ubiquinone oxidoreductase (complex I) (7)(8)(9)(10)(11)(12)(13)(14)(15). These LHON-associated mtDNA mutations often occur near homoplasmy or at homoplasmy. These LHON-associated mtDNA mutations conferred mild mitochondrial dysfunctions, especially the reduced activity of complex I (16)(17)(18)(19). Notably, families carrying the mtDNA mutation(s) exhibited gender bias and incomplete penetrance of optic neuropathy (20,21). These indicated that nuclear and mitochondrial genetic modifiers are required for the phenotypic manifestation of these LHON-associated mtDNA mutations. In a recent study, we identified two nuclear modifiers (c.572G.T, p.Gly191Val in YARS2; and c.157C.T, p.Arg53Trp in PRICKLE3) necessary for the phenotypic expression of the m.11778G.A mutation (22,23). Several mtDNA variants, including tRNA Met 4435A.G, tRNA Thr 15951A.G, ND1 3394T.C, and ND6 14502T.C mutations, may act as mitochondrial genetic modifiers to increase the phenotypic expression of the m.11778G.A mutation (24)(25)(26)(27)(28). Furthermore, our previous investigations identified the m.3866T.C mutation in 7 probands (4 bearing only the m.3866T.C mutation and 3 carrying m.3866T.C and m.11778G.A mutations) in a large cohort of 1281 Chinese probands with LHON (15,29). These Chinese pedigrees carrying only the m.3866T.C mutation exhibited extremely low penetrance of LHON, whereas the penetrances of LHON among those pedigrees bearing the m.3866T.C mutation together with m.11778G.A mutations were higher than those in families carrying only single mtDNA mutation (15,29). These data suggested the role of m.3866T.C mutation in the pathogenesis of LHON. However, the pathogenic mechanisms underlying these LHONassociated mutations remain elusive.
The m.3866T.C mutation resulted in the change of highly conserved isoleucine at position 187 with threonine (I187T) in  (19,34). Using Western blotting and blue native PAGE (BN-PAGE) analyses, we examined if the m.3866T.C and m.11778G.A mutations exerted an effect on the stability of ND1 and ND4 as well as assembly of complex I. These cell lines were then assessed for effects of the mtDNA mutations on the enzymatic activities of respiratory chain complexes, the rate of O 2 consumption, mitochondrial ATP production, mitochondrial membrane potential, and generation of reactive oxygen species (ROS). These cell lines were further evaluated for the effect of these mtDNA mutations on apoptotic state and mitophagy.

Results
Clinical and genetic evaluation of 10 Chinese families carrying the m.3866T>C mutation Ten Han Chinese pedigrees bearing the m.3866T.C mutation (6 bearing only the m.3866T.C mutation, 4 carrying m.3866T.C mutation together with m.11778G.A mutation) were identified in a large cohort of 1793 Chinese probands with LHON (15). This translated to an incidence of 0.6% in this cohort. All available members of 10 pedigrees underwent comprehensive physical and ophthalmologic examinations to identify personal or family medical histories of visual impairments and other clinical abnormalities. As shown in Table S1 and Fig. S1, matrilineal relatives exhibited variable penetrance and expressivity of optic neuropathy among and within families. In particular, the severity of visual loss ranged from profound visual loss to normal vision. The age at onset of optic neuropathy of 6 pedigrees bearing only m.3866T.C mutation ranged from 5 to 39 years, with an average of 24.8 years, whereas the average age at onset of visual impairment of 4 pedigrees harboring both m.3866T.C and m.11778G.A mutations were 22.2 years, respectively. Strikingly, 6 pedigrees harboring only the m.3866T.C mutation exhibited extremely low penetrance of optic neuropathy, ranging from 11.1% to 29.2%, with an average of 17.0%. In contrast, the average penetrances of optic neuropathy of 4 pedigrees harboring both m.3866T.C and m.11778G.A mutations were 41.9%.
Entire sequence analysis of mtDNAs among these Chinese pedigrees revealed that these probands showed the presence of m.3866T.C or m.11778G.A mutations and distinct sets of mtDNA polymorphisms, including 129 known variants, as shown in Table S2. As shown in Table S1, the mtDNAs from 10 pedigrees resided at mtDNA haplogroups C4a (1), D4 (2), G2 (1), M 10 (4), H2 (1), and R (1), respectively (35). These mtDNA variants included 35 in the D-loop region, 4 in the 12S rRNA gene, 3 in the 16S rRNA gene, 5 in the tRNA gene, and 53 silent and 29 missense variants in the genes encoding polypeptides (36). These variants in RNAs and polypeptides were evaluated by phylogenetic analysis of these variants and sequences from 17 vertebrates, as shown in Table S2. These variants were further evaluated for the presence of 485 control subjects and potential structural and functional alterations. Of these, these variants may not have potential structural and functional alterations. These findings indicated that these mtDNA variants do not play an important role in the phenotypic manifestation of m.3866T.C mutations.
The effect of m.3866T>C and m.11778G>A mutations on stability of ND1 and ND4 To test the effect of m.3866T.C (I187T) and m.11778G.A (R340H) mutations on the structure and function of ND1 and ND4, we performed the tertiary structure analysis of complex I from H. sapiens and M. musculus. Based on the Cryo-EM structure of mammalian complex I (PDB entry 5XTD and 6G2J) (32,33), the hydrophobic side chain of I187 forms the hydrophobic interaction with phosphatidylcholine in ND1 (Fig. 1A), whereas the side chain guanidino group of R340 forms specific electrostatic interactions with the backbone hydroxyl group of V221 in ND4, which is located at the matrix side and water-membrane interface (Fig. 1B). Hence, the replacement of hydrophobic isoleucine at position 187 with the hydrophilic threonine by m.3866T.C mutation in ND1 or arginine at position 340 with histidine by m.11778G.A mutation in ND4 may destabilize these interactions inside ND1 or ND4, thereby perturbing the structure and stability of ND1 and ND4 proteins, respectively.
To experimentally test this hypothesis, we examined the levels of ND1 and ND4 proteins by Western blotting in these mutant cell lines carrying only m.3866T.C mutation, only m.11778G.A mutation, or both m.3866T.C and m.11778G.A mutations, as well as control cell lines lacking these mutations. As shown in Fig. 1 The complex I of human and mice is composed of 45 subunits, including 7 subunits encoded by mtDNA and 38 subunits encoded by nuclear genes. These subunits are organized into Synergy effect of two LHON-associated mtDNA mutations six modules, Q, ND1, ND2, ND4, ND5, and N, with the help of specific assembly factors that begin to assemble separately and ultimately combine to form the complex I holocomplex in the inner membrane of mitochondria (37). As shown in Fig.  2A, ND1 interacts with NDUFA1, NDUFA3, NDUFA8, and NDUFA13, whereas ND4 interacts with ND5, NDUFS2, NDUFB1, NDUFB4, NDUFB5, NDUFB8, and NDUFB11 (32,33). To examine whether the m.3866T.C or m.11778G.A mutation affected the expression of other subunits of complex I, we measured the levels of NDUFA3, NDUFA8, NDUFA13, ND5, NDUFS2, NDUFB4, NDUFB5, NDUFB8, and NDUFB11 by Western blot analysis among mutant and control cell lines, respectively. As shown in Fig. 1, C and D, and 2, B-D, the levels of these subunits in mutant cell lines were comparable with those in control cell lines. We then assessed the consequences of this m.3866T.C mutation on complex I activity using isolated mitochondrial from our mutant and control cybrid cell lines (40)(41)(42). As shown in Fig. 3C, the activities of complex I in mutant cell lines carrying only m.3866T.C, only m.11778G.A, and both m.3866T.C and m.11778G.A mutations were 76.7%, 59.8%, and 53.4%, respectively, relative to the average values of control cell lines. As shown in Fig. 3, D-F, the average activities of complexes II, III, and IV in mutant cell lines were comparable with those of the control cell lines.

Respiration deficiency
To further examine whether the m.3866T.C mutation affected cellular bioenergetics, we examined the oxygen consumption rates (OCR) of various mutant and control cybrid cell . Based on the cryo-EM structure of complex I from H. sapiens (PDB entry 5XTD), the hydrophobic residue I187 interacts with a PC in ND1. I187 and PC are highlighted in red and magenta sticks, respectively. The other hydrophobic residues around I187 are shown in a yellow line. The distance between the side chain of I187 and the fatty acyl tail of PC is shown in a black number (in angstrom). B, the electrostatic interactions in R340 of ND4 (shown in blue). Based on the cryo-EM structure of complex I from M. musculus (PDB entry 6G2J), the side chain guanidino group of R340 forms specific electrostatic interactions with the backbone hydroxyl group of V221 in MT-ND4. R340 and its partner are highlighted in red and white sticks, respectively. Hydrogen bonds are shown as black-dashed lines, with the distance shown in black numbers (in angstroms). C, Western blot analysis. 5 mg of mitochondrial proteins from various cell lines was electrophoresed through a denaturing polyacrylamide gel, electroblotted, and hybridized with ND1, N4, and ND6 antibodies, respectively, with VDAC as a loading control. D, quantification of ND1, ND4, and ND6 in mutant and control cell lines. The calculations were based on three independent determinations in each cell line. The error bars indicate two standard errors of the means. p indicates the significance, according to the t test, of the differences between mutant and control cell lines.
lines with a Seahorse Bioscience XF-96 extracellular flux analyzer (Seahorse Bioscience) (43,44). As shown in Fig. 4, the basal OCR in the mutant cybrids carrying only m.3866T.C, only m.11778G.A, or both m.3866T.C and m.11778G.A mutations were 82.9%, 79.1%, and 61.0%, respectively, relative to the mean value measured in the control cybrids. To assess which of the enzyme complexes of the respiratory chain was perturbed in the mutant cell lines, OCR was measured after the sequential addition of oligomycin (inhibit the ATP synthase), carbonyl cyanide p-(trifluoromethoxy) phenylhydrazone (FCCP; to uncouple the mitochondrial inner membrane and allow for maximum electron flux through the electron transport chain), rotenone (to inhibit complex I), and antimycin (to inhibit complex III). The difference between the basal OCR and the drug-insensitive OCR yielded an amount of ATP-linked OCR, proton leak OCR, maximal OCR, reserve capacity, and nonmitochondrial OCR. As shown in Fig. 4

Reductions in mitochondrial ATP production
To examine the capacity of oxidative phosphorylation, we measured the levels of cellular and mitochondrial ATP production using a luciferin/luciferase assay (44). Populations of cells were incubated in the media in the presence of glucose and 2-deoxy-D-glucose with pyruvate to inhibit the glycolysis. As shown in Fig. 5, the levels of mitochondrial ATP production (pyruvate and 2-deoxy-D-glucose) in the mutant cybrids bearing only m.3866T.C, only m.11778G.A, and both m.3866T.C and m.11778G.A mutations were 82.3%, 67.2%, and 43.2%, respectively, of average values of control cybrids, whereas the levels of cellular ATP production (glucose) in mutant cybrids were comparable with those measured in the control cybrids.

Alterations in mitochondrial membrane potential
The mitochondrial membrane potential (DC m ) generated by proton pumps (complexes I, III, and IV) is an essential component in the process of energy storage during oxidative phosphorylation (45,46). We examined the levels of mitochondrial membrane potential (DC m ) in the mutant and control cybrids using a fluorescence probe JC-10 assay system. As shown in Fig. 6, the levels of the DC m in the mutant cybrids carrying only the m.3866T.C mutation, only the m.11778G.A mutation, or both the m.3866T.C and m.11778G.A mutations were 79.9%, 75.7%, and 53.7% of the mean values measured in the control cell lines. Conversely, the levels of DC m in mutant cell lines in the presence of FCCP were comparable with those measured in the control cell lines.

The increase of ROS production
Mitochondrial ROS play a critical role in physiological consequences (47). We next assessed ROS production in mutant cybrid cell lines via flow cytometry, comparing baseline staining intensity for each cell line to that upon oxidative stress to obtain a ratio corresponding to ROS generation (48,49). Geometric mean intensity was recorded to measure the rate of mitoROS of each sample. The relative levels of geometric mean intensity in each cell line were calculated to delineate the levels of mitoROS in mutant and control cells. As shown in Fig. 7, A and B, the levels of ROS generation in the mutant cybrids carrying only the m.3866T.C mutation, only the m.11778G.A mutation, and both the m.11778G.A and m.3866T.C mutations were 119.3%, 122.1%, and 141.2% of the mean values measured in the control cell lines. Furthermore, we examined the levels of catalase and superoxide dismutase proteins (SOD1 and SOD2) in mutant and control cell lines by Western blot analysis (47,50). As showed in Fig. 7, C and D, the marked increasing levels of these proteins were observed in the mutant cybrids. In par-  5 mM), carbonyl cyanide p-(trifluoromethoxy) phenylhydrazone (FCCP) (0.5 mM), rotenone (1 mM), and antimycin A (1 mM) at the indicated times to determine different parameters of mitochondrial functions. B, graphs present the ATP-linked OCR, proton leak OCR, maximal OCR, reserve capacity, and nonmitochondrial OCR in mutant and control cell lines. Nonmitochondrial OCR was determined as the OCR after rotenone/antimycin A treatment. Basal OCR was determined as OCR before oligomycin minus OCR after rotenone/antimycin A. ATP-lined OCR was determined as OCR before oligomycin minus OCR after oligomycin. Proton leak was determined as basal OCR minus ATP-linked OCR. Maximal OCR was determined as the OCR after FCCP minus nonmitochondrial OCR. Reserve capacity was defined as the difference between maximal OCR after FCCP minus basal OCR.

Promoting apoptosis
Deficient activities of oxidative phosphorylation have been linked to protection against certain apoptotic stimuli (19,51).
To evaluate if the m.3866T.C and m.11778G.A mutations affected the apoptotic processes, we examined the apoptotic state of mutant and control cybrid cell lines by using Annexin V/PI-based flow cytometry and Western blot analyses. As shown in Fig. 8A, the average ratio of Annexin V-positive cells in the mutant cybrids carrying only the m.3866T.C mutation, only the m.11778G.A mutation, and both m.11778G.A and m.3866T.C mutations were 223.9%, 276.6%, and 330.4% of the mean values measured in control cell lines. Furthermore, we examined the levels of three apoptosis-activated proteins (caspase 9, caspase 3, and poly-ADP-ribose polymerase [PARP]) in mutant and control cell lines by Western blot analysis (52). As shown in Fig. 8B, the markedly increasing levels of these proteins were observed in the mutant cybrids. In particular, the levels of cytochrome c, caspase 9, caspase 3, and PARP in mutant cybrids carrying both m.3866T.C and m.11778G.A mutations were 218.2%, 166.1%, 135.7%, and 193.4%, those in cybrids carrying only the m.3866T.C mutation were 125.5%, 143.5%, 122.4%, and 134.5%, and those in cybrids carrying only the m.11778G.A mutation were 185.7%, 154.5%, 127.9%, and 174.5% relative to the mean values measured in the control cybrids, respectively (Fig. 8, C and D).

Alteration in mitophagy
The alterations in oxidative phosphorylation (OXPHOS) and mitochondrial membrane potential affected the mitophagic removal of damaged mitochondria (19,53,54). To investigate if the m.3866T.C and m.11778G.A mutations regulated the autophagy, the autophagic states of various mutant and control cybrids were analyzed using both fluorescence-based cytometry and Western blotting assays. First, CYTO-ID ® autophagy detection kits were used with flow cytometry to examine the degree of autophagy of mutant and control cell lines. As shown in Fig. 9, A and B, significant shifts in the fluorescence peak to high intensity occurred in the mutant cybrids compared with those in control cybrids. The levels of autophagy in the mutant cybrids harboring only the m.3866T.C mutation, only the m.11778G.A mutation, and both m.11778G.A and m.3866T.C mutations were 82.3%, 65.5%, and 55.4% of the mean values measured in the control cybrids. To further evaluate the effect of mtDNA mutations on mitophagy, we performed a Western blot analysis using two markers: microtubule-associated protein 1A/1B light chain 3B (LC3) and sequestosome 1 (SQSTM1/p62) in various cell lines. LC3-II is recleaved by cysteine protease (Atg4B) following completion of the autophagosome and recycled, whereas SQSTM1/p62, one of the best-known autophagic substrates, interacts with LC3 to ensure the selective delivery of these proteins into the autophagosome (55). As shown in Fig. 9

Discussion
The pathogenicity of ND1 3866T>C mutation In this present study, we further investigated the molecular mechanism of LHON-associated ND1 3866T.C mutation. The m.3866T.C mutation has been associated with LHON, either together with the m.11778G.A mutation or in isolation among Chinese families (15,29). The incidences of the m.3866T.C mutation were 0.6% in the cohorts of 1793 Chinese probands with LHON. The occurrence of the m.3866T.C mutation in these genetically unrelated pedigrees affected by LHON and differing considerably in their mtDNA haplotypes (C4a, D4, G2, M10, H2, and R) strongly indicated that this mutation is involved in the pathogenesis of LHON (15,29). However, 5 pedigrees carrying only m.3866T.C did not have a family history of optic neuropathy. The extremely lower penetrance of LHON in these families carrying only the m.3866T.C mutation was comparable with those in Chinese families bearing only one LHON-associated mtDNA mutation (s), including m.3394T.C, m.14502T.C, m.12338T.C, m.11778G.A, or m.14484T.C (26,27,(56)(57)(58)(59).
The m.3866T.C (I187T) mutation changed a conserved tyrosine at position 187 with isoleucine in ND1, which is an essential subunit of complex I (29, 60). Complex I is composed of 45 subunits, including 7 subunits encoded by mtDNA and 38 subunits encoded by nuclear genes. These subunits are organized into six modules, Q, ND1, ND2, ND4, ND5, and N, with the help of specific assembly factors that begin to assemble separately and ultimately combine to form the complex I holocomplex in the inner membrane of mitochondria (30)(31)(32). Based on the cryo-EM structure of mammalian complex I, the hydrophobic side chain of I187 forms the hydrophobic interaction with lipids in ND1. The substitution of hydrophobic isoleucine at position 187 for the hydrophilic threonine by the m.3866T.C mutation in ND1 may destabilize these interactions, thereby altering the structure and function of complex I. In this investigation, the instability of mutated ND1 was evidenced by the reduced levels of ND1 observed in cybrid cell lines bearing only the m.3866T.C mutation and both m.3866T.C and m.11778G.A mutations. However, the m.3866T.C mutation did not affect the expression of other subunits of complex I, such as NDUFA3, NDUFA8, and NDUFA13, which interacted with ND1. The mutated ND1 indeed perturbed the assembly of complex I, as in the case of m.14502T.C or m.3394T.C together with m.11778G.A mutations (26,27). Both instability of ND1 and altered assembly of complex I were responsible for 20% reductions in the activity of complex I observed in cybrids carrying only the m.3866T.C mutation. Furthermore, the m.3866T.C mutation yielded reduced rates in the basal OCR and ATP-linked OCR in mutant cybrid cell lines. These respiratory deficiencies would lead to an impaired ATP synthesis, alterations in the mitochondrial membrane potentials, and increasing production of oxidative reactive species (61,62). In the present study, there was a very significant correlation between the reduced levels of complex I activity, mitochondrial ATP production, mitochondrial membrane potentials, and the increasing rates of ROS production (p , 0.01) in the control and mutant cybrids harboring the m.3866T.C mutation. This correlation is clearly consistent with the important role that the defective activity of complex I in the mutant cell lines carrying m.3866T.C mutation plays in producing mitochondrial dysfunction for the development of LHON.
Mitochondrial dysfunctions and overproduction of ROS by LHON-associated mtDNA mutations often induced mitochondrion-dependent apoptotic death and defective mitophagy (19,62,63). In this study, the m.3866T.C mutation promoted the apoptotic process, evidenced by elevated release of cytochrome c into cytosol and increased levels of apoptosis-activated proteins: caspase 9, caspase 3, and poly-ADP-ribose polymerase in the cybrid cell lines carrying the m.3866T.C mutation. Moreover, we also documented the involvement of m.3866T.C mutation in decreased mitophagy, as shown by reduced levels of autophagy protein LC3 and accumulation of autophagic substrate p62, as in the cases of other LHON-linked mtDNA mutations (19,53,54). These data demonstrated the direct link between mitochondrial dysfunction caused by complex I mutation and apoptosis or mitophagy. However, the extremely low penetrance of LHON and mild mitochondrial dysfunction strongly suggested that the m.3866T.C mutation is the primary factor underlying the development of LHON but is insufficient to produce a clinical phenotype. The additional hereditary or environmental factors are necessary for the phenotypic expression of m.3866T.C mutation.
The synergy effect of m.3866T>C mutation on the phenotypic expression of the LHON-associated m.11778G>A mutation Mitochondrial genetic backgrounds, such as specific mtDNA variants, may act as genetic modifiers to increase the phenotypic expression of the m.11778G.A mutation (24)(25)(26)(27)(28)64). The average penetrances of optic neuropathy of 4 pedigrees harboring both m.3866T.C and m.11778G.A mutations exhibited levels significantly higher than those in 6 families carrying only the m.3866T.C mutation and other pedigrees harboring only the m.11778G.A mutation. (14,15,29,58). These observations strongly suggested that the m.3866T.C mutation increases the expression of optic neuropathy in these families carrying the m.11778G.A mutation, as in the cases of m.4216T.C and m.13708G.A mutations in European families and m.4435A.G, m.3394T.C, m.15951A.G, and m.14502T.C mutations in other Chinese pedigrees (11,(24)(25)(26)(27)64).
These cybrid cell models, with identical nuclear background and mtDNA haplogroup, allow us to evaluate the biochemical consequences because of specific mtDNA mutations. In fact, m.11778G.A only altered the stability of ND4 but not other subunits, such as ND5, NDUFS2, NDUFB4, NDUFB5, NDUFB8, and NDUFB11, that interacted with ND4. As a result, m.11778G.A mutation conferred a mild reduction in the activity of complex I (16-18, 22, 23). In the present study, cybrid cell lines bearing both m.3866T.C and m.11778G.A mutations exhibited more severe mitochondrial dysfunctions than those in the cells carrying only the m.11778G.A or m.3866T.C mutation. In particular, the cell lines carrying both m.3866T.C and m.11778G.A mutations exhibited more unstable complex I than those bearing only the m.3866T.C or m.11778G.A mutation. As a result, the m.3866T.C mutation aggravated the biochemical defects caused by the m.11778G.A mutation, including the defective activity of complex I, decreases of mitochondrial ATP production and mitochondrial membrane potentials, and increasing production of ROS. In particular, the 57% decreases in the mitochondrial ATP production observed in the cybrids bearing both m.3866T.C and m.11778G.A mutations were below a proposed threshold level, which is 50% of control levels in the mitochondrial ATP, to develop a clinical phenotype (23,65,66). In particular, RGCs require the highest-ATP-demand cells in the retina (67,68). Therefore, the retinal ganglion cells carrying both m.3866T.C and m.11778G.A mutations may be preferentially involved, because they are high-ATP-demand cells and somehow are exquisitely sensitive to subtle imbalance in cellular redox state or increased level of free radicals (1,5,6,(69)(70)(71). Mitochondrial dysfunction affected the apoptotic sensitivity and mitophagy of cells carrying the LHON-associated mtDNA mutations (19,53,71). In the present investigation, the cell lines carrying both m.3866T.C and m.11778G.A mutations exhibited more apoptotic susceptibility than those carrying only the m.3866T.C or m.11778G.A mutation. In particular, cell lines carrying both m.3866T.C and m.11778G.A mutations exhibited an increase of apoptosis and higher levels of apoptosis-activated proteins cytochrome c, caspases 3 and 9, and PARP than those in cell lines carrying only the m.3866T.C or m.11778G.A mutation. These data demonstrated that m.3866T.C mutation elevated the apoptosis induced by the m.11778G.A mutation. Mitophagy disposes of damaged mitochondria and maintains the healthy mitochondrial population in the cells. The impairment of OXPHOS and alteration of mitochondrial membrane potential affected the mitophagic removal of damaged cells (72,73). In this study, the cell lines carrying both m.3866T.C and m.11778G.A mutations exhibited more increased levels of p62 than cells carrying only m.3866T.C or m.11778G.A mutation, which indicated the preferential accumulation of autophagic substrates, such as misfolded proteins, leading to deleterious effects. Furthermore, more severely decreased levels of LC3 were observed in cybrid cell lines carrying both m.3866T.C and m.11778G.A mutations than those in cell lines carrying only m.3866T.C or m.11778G.A mutation, which suggested more decreases in the capacity of the double mutant cells to generate autophagosomes, thereby perturbing the autophagic degradation of ubiquitinated proteins. These data indicated that m.3866T.C mutation worsened the defective mitophagy Figure 9. Analysis of mitophagy. A, flow cytometry histograms showing fluorescence of 6 mutant and 2 control cybrids using a CYTOID® autophagy detection kit. Cells were incubated with DMEM in the absence and presence of rapamycin (inducers of autophagy) and chloroquine (lysosomal inhibitor) at 37°C for 18 h, added to CYTO-ID ® -Green dye, and analyzed using a Novocyte flow cytometer (ACEA Biosciences). B, relative fluorescence intensity from various cell lines. Three independent determinations were done in each cell line. C, Western blot analysis for mitophagic response proteins LC3-I/(I1II) and P62. Twenty micrograms of total cellular proteins from various cell lines was electrophoresed, electroblotted, and hybridized with LC3 and p62, with GAPDH as a loading control. D, quantification of autophagy markers LC3 I/II and p62 in mutant and control cell lines, determined as described elsewhere (19). Four independent determinations were done in each cell line. Graph details and symbols are explained in the legend to Fig. 1. caused by the m.11778G.A mutation. The significant correlation between severe biochemical phenotypes in these mutants cybrids and higher penetrance of LHON in these pedigrees harboring both m.11778G.A and m.3866T.C mutation demonstrated that the m.3866T.C mutation enhanced the penetrance and occurrence of optic neuropathy in the Chinese families carrying both m.3866T.C and m.11778G.A mutations. Therefore, our findings highlighted the critical role of m.3866T.C mutation in the pathogenesis of LHON, manifested by sole mtDNA mutation or its synergy with the m.11778G.A mutation.

Families and subjects
A total of 1793 genetically unrelated Han Chinese subjects with LHON were recruited from eye clinics across China, as described previously (14,15,26). This study was in compliance with the Declaration of Helsinki. Informed consent, blood samples, and clinical evaluations were obtained from all participating family members under protocols approved by the Ethic Committees of Zhejiang University and Wenzhou Medical University. A comprehensive history and physical examination for these participating subjects were performed at length to identify both personal or family medical histories of visual impairment and other clinical abnormalities. The ophthalmic examinations of probands and other members of these families were conducted as detailed previously (24,29). A total of 485 control DNA samples were obtained from adult Han Chinese from the same area.

Mitochondrial DNA analysis
The mutational screening of the m.3866T.C mutation was performed as detailed previously (15,29). The entire mitochondrial genomes of 6 probands carrying only the m.3866T.C mutation and 4 individuals harboring both the m.3866T.C and m.11778G.A mutations were determined as detailed elsewhere (74). The resulting sequence data were compared with the updated consensus Cambridge sequence (GenBank accession number NC_012920) (36). The entire mtDNA sequences of 11 pedigrees and 1 control subject were assigned to the mitochondrial haplogroups using the nomenclature described previously (35). An analysis for the presence and levels of m.11778G.A and m.3866T.C mutations in mutant and control cell lines were carried out as described previously (23,29). The quantification of mtDNA copy numbers from different cell lines was carried out as detailed elsewhere (73).

Cell lines and culture conditions
Lymphoblastoid cell lines derived from three probands (WZ77-III5 harboring only m.3866T.C mutation, WZ455-III9 bearing only m.11778G.A mutation, and WZ449-IV5 carrying both m.11778G.A and m.3866T.C mutations) and one control individual, WZC46, were immortalized by transformation with the Epstein-Barr virus, as described elsewhere (75). Lymphoblastoid cell lines were grown in RPMI 1640 medium (Corning) supplemented with 10% fetal bovine serum. The 143B.TKcell line was grown in DMEM (containing 4.5 mg of glucose and 0.11 mg pyruvate per ml), supplemented with 100 mg of BrdU per ml and 5% FBS. The mtDNA-less r o 206 cell line, derived from 143B.TK -, was grown under the same conditions as the parental line, except for the addition of 50 mg of uridine/ml (34). Transformation by cytoplasts of mtDNA-less r o 206 cells was performed by using four immortalized lymphoblastoid cell lines, as detailed previously (34,76). The cybrids derived from each donor cell line were analyzed for the presence and level of the m.3866T.C or m.11778G.A mutation and mtDNA copy numbers as detailed elsewhere (24,29,76). Two cybrids derived from each donor cell line with homoplasmy of mtDNA mutations and similar mtDNA copy numbers were used for the following biochemical characterization. All cybrid cell lines were maintained in the same medium as the 143B.TKcell line.

Western blot analysis
Mitochondria were isolated from mutant and control cell lines by following the protocol described elsewhere (77). Western blot analysis was carried out by using total mitochondrial proteins isolated from mutant and control cell lines, as detailed elsewhere (44,46 ). The protein signals were detected using the ECL system (Millipore). Quantification of density in each band was performed as detailed previously (22,46).

BN gel electrophoresis
BN gel electrophoresis and in-gel activity assay were performed by using mitochondrial proteins, as detailed elsewhere (27,38,39).

Measurements of oxygen consumption
The rates of oxygen consumption in cybrid cell lines were measured with a Seahorse Bioscience XF-96 extracellular flux analyzer (Seahorse Bioscience), as described elsewhere (43).
Synergy effect of two LHON-associated mtDNA mutations ATP measurements The cellular and mitochondrial ATP levels were analyzed by the Cell Titer-Glo luminescent cell viability assay kit (Promega) according to the modified procedures of the manufacturer (44).

Assessment of mitochondrial membrane potential
Basal level of mitochondrial membrane potential was examined with JC-10 assay kit-flow cytometry (Abcam) according to the manufacturer's general recommendations, with some modifications, as detailed elsewhere (45,46).

ROS measurements
The MitoSOX red mitochondrial superoxide indicator (Thermo Fisher) was used for ROS measurements by following the manufacturer's instructions, as detailed previously (48,49).

Annexin V/PI apoptosis assay by flow cytometry
For discrimination of apoptotic and nonapoptotic cells by Annexin V/PI staining, cells were harvested and stained with Annexin V and 1 ml of propidium iodide (PI) (V13242, Ther-moFisher Scientific) according to the manufacturer's instructions. Each sample was detected by NovoCyte (ACEA Biosciences) and analyzed using NovoExpress software (79).

Analysis of autophagy
The fluorescence-based cytometry to analyze the level of mitophagy was performed using the CYTO-ID ® autophagy detection kit (Enzo) as detailed elsewhere (80,81).

Statistical analysis
All statistical analyses were performed using the unpaired, two-tailed Student's t test contained in the GraphPad Prism 5 program (GraphPad Software) and Microsoft Excel (version 2016). A p value of ,0.05 was considered statistically significant.

Data availability
Representative experiments are shown in the figures and supporting information. Supplemental data include two tables and one figure. For any additional information, please contact the corresponding author.