Mitochondrial replacement in an iPSC model of Leber's hereditary optic neuropathy

Cybrid technology was used to replace Leber hereditary optic neuropathy (LHON) causing mitochondrial DNA (mtDNA) mutations from patient-specific fibroblasts with wildtype mtDNA, and mutation-free induced pluripotent stem cells (iPSCs) were generated subsequently. Retinal ganglion cell (RGC) differentiation demonstrates increased cell death in LHON-RGCs and can be rescued in cybrid corrected RGCs.

AGING that encode for the mitochondrial Complex I subunits [9,10]. These homoplasmic mutations are shown to disrupt the activity of Complex I, leading to a decrease in bioenergetic production and an increased level of oxidative stress [11]. However, the precise mechanism for disease progression in LHON remains unknown. Here we report the use of iPSCs to model LHON, and demonstrate generation of isogenic iPSC controls by replacing LHON mtDNA using cybrid technology.

RESULTS
We previously reported on the generation and characterisation of LHON iPSCs [12]. For this study, we utilised iPSCs from a healthy control (MRU11780), and a LHON patient (LHON Q1-4) with homoplasmic double mtDNA mutations m.4160T>C and m.14484T>C which affected the MT-ND1 and MT-ND6 genes respectively [12]. This LHON patient exhibited AGING "LHON plus" phenotype, with clinical features including optic nerve atrophy, juvenile encephalopathy and peripheral neuropathy [13]. To generate an isogenic control for iPSC modeling, we utilised the cybrid technique to replace the mutant mtDNA in LHON fibroblasts. LHON fibroblasts were pre-treated with rhodamine 6-G to disable the transmission of endogenous mtDNA, followed by fusion with donor mitochondria obtained from wild-type keratinocytes (Fig. 1A). After 27 days post-fusion, proliferating fibroblast colonies that are indicative of successful mitochondrial replacement were isolated and expanded (Fig. 1B). In contrast, no proliferating fibroblast colony was observed in control conditions that did not receive donor keratinocyte mitochondria (Fig. 1B). Out of 12 clones screened, we identified 1 cybrid clone with the corrected mtDNA genotype at m.4160 and m.14484 (Fig. 1C). Microsatellite analysis of a panel of 12 poly-morphic markers confirmed that the corrected cybrid clone originated from the parental LHON fibroblasts, whereas the donor keratinocytes possessed a different microsatellite profile (Fig. 1D, Supplementary Table). We then generated iPSCs from the cybrid fibroblasts using the episomal method. Following reprogramming, three clones of cybrid iPSCs (CYB iPSC c1, CYB iPSC c2, CYB iPSC c3) were selected for this study. Importantly, all cybrid iPSC clones retained mtDNA correction with no detectable mutations at m.4160 and m.14484 (Fig. 1C, Supplementary Fig. 1). Further characterization demonstrated that the cybrid iPSCs expressed the pluripotent markers OCT-4 and TRA-1-60 ( Fig. 1E, Supplementary Fig. 2). The derived cybrid iPSCs were also differentiated into cells of the three germ layers in vitro by embryoid body formation and in vivo by teratoma formation (Fig. 1F  Data are expressed as mean + SEM of independent samples (n=11 for control, n=8 for LHON and n=8 for corrected cybrid, expressed as pooled data of experimental repeats and biological repeats (3 clones)) (B) TUNEL assay revealed increased susceptibility to apoptosis in LHON RGCs and reversal in corrected cybrid lines. Data are expressed as mean of each clone, n = 3 clones, error bars = mean ± SEM. (C) Quantification of mitochondrial superoxide levels using MitoSOX in control, LHON and corrected cybrid RGCs. Error bars = ± SEM, n = 3 clones. Statistical significance was established by one way ANOVA followed by Dunnett's test for multiple comparisons, ** p<0.01, * p<0.05, ns: not significant.
(B) Representative images of control fibroblasts (no fusion) and fused fibroblasts that received donor mitochondria (cybrid fusion) at 8, 11, 27 days post R6G treatment. (C) Genotype confirming cybrid correction of mutation in fibroblasts and the corresponding iPSCs. Red arrows indicate LHON mutations at m.4160T>C and m.14484T>C, blue arrows indicate wild-type genotype. Note that the genotype of the parental LHON fibroblasts (LHON Q1-4) was reported previously [12]. ed no chromosomal abnormalities in the derived cybrid iPSC clones (Fig. 1G, Supplementary Fig. 4). Together, these results demonstrate the feasibility of using the cybrid technique to generate isogenic iPSC controls for mtDNA disease modeling.
Finally, we assessed the effect of the LHON mtDNA mutations in iPSC-derived RGCs. Control, LHON and cybrid iPSCs were directed to differentiate into RGCs by a stepwise differentiation method that we recently published, which demonstrated an enriched population of functional RGCs [14]. Three iPSC clones per patient were used and all clones were able to differentiate into RGCs with similar efficiency (Fig. 2A). Following RGC enrichment using MACS, TUNEL analysis revealed an increased level of apoptosis in LHON RGCs, from 13.6 ± 1.9% in control RGCs to 56.1 ± 6.8% in LHON RGCs (Fig. 2B). Importantly, this effect was reverted in the cybrid corrected RGCs, with the apoptosis level returned to control levels (Fig. 2B, 12.9 ± 4.4%), demonstrating that the increased susceptibility to cell death observed in LHON-RGCs is a direct consequence of LHON mtDNA mutations. Mitochondrial dysfunction in RGCs was then assessed using MitoSOX, which measures the levels of mitochondrial superoxide, as an indication of mitochondrial oxidative stress. Despite a trend of an increased superoxide levels in LHON RGCs compared to control RGCs, this difference in mitochondrial superoxide levels was not statistically significant, possibly due to high variations observed amongst the iPSC clones. Lower superoxide levels were also observed in corrected cybrid lines compared to LHON RGCs (Fig. 2C). Together these results suggest that the higher apoptosis observed in the LHON RGCs cannot be explained by elevated mitochondrial superoxide alone. Future studies to investigate other defects in LHON RGCs, such as ATP deficiency [15] or mitochondrial biogenesis [16], will help elucidate the mechanism underlying penetrance and disease progression of LHON.

DISCUSSION
The recent development of mitochondrial replacement therapy using pronuclear transfer offers exciting prospects to correct mtDNA mutations in the early human embryo [17]. However, pronuclear transfer is technically challenging, especially in cells of small size, and requires specialised equipment. Here, we describe for the first time a cybrid approach to correct mtDNA mutations in an iPSC disease model. Compared to pronuclear transfer, the cybrid technique is easier to perform and can be adapted to generate an isogenic control for iPSC models of homoplasmic mtDNA diseases. Reduction of mutant mtDNA load, rather than complete replacement of mutant mtDNA with wild-type mtDNA, might be sufficient for phenotypic rescue in iPSC isogenic controls. In support of this, previous studies have shown that selective elimination of mutant mtDNA by mitochondria-targeted TALEN can reduce the mutant mtDNA loads and restore the mitochondrial dysfunction [18]. Very recently, zinc fingers were also used for the successful replacement of heteroplasmic mtDNA [19]. Our results show an increased susceptibility to apoptosis in LHON iPSC-derived RGCs. Importantly, apoptosis was returned to normal levels in the cybrid-corrected RGCs, hence demonstrating that the increased susceptibility to cell death observed in RGCs is a direct consequence of LHON mtDNA mutations.
In summary, our approach shows the advantage of using LHON patient-derived iPSCs and their isogenic cybrid control as a platform for the study of the fundamental mechanisms underlying LHON pathogenesis. Moreover, the cybrid technique provides a feasible strategy to correct mtDNA mutations in iPSC models, which could be applied to model other mtDNA diseases.

Ethics
All experimental work performed in this study was approved by the Human Research Ethics Committees of the Royal Victorian Eye and Ear Hospital (11/1031H) and the University of Melbourne (0605017) [20,21], and with the Animal Ethics Committee of St Vincent's Hospital (002/14), in accordance with the requirements of the National Health & Medical Research Council of Australia and conformed with the Declarations of Helsinki.

Cybrid generation
Cybrid transfer was performed as described [22,23]. Briefly, human epidermal keratinocytes (System Bioscience) were used as donor by enucleation using cytochalasin B treatment and high speed centrifugation. LHON Q1-4 fibroblasts were pre-treated with 2.5µg/ml rhodamine 6-G for 5 days, before fusion with donor cell cytoplasts using polyethylene glycol. On the next day, fused cells were replated and allowed to culture for up to 32 days. Proliferating colonies were picked and expanded using cloning cylinders (Corning). Isolated fibroblast clones were genotyped for mtDNA mutations at mt.4160 and mt.14484 to screen for successful mitochondrial replacement.

iPSC generation and characterisation
Reprogramming of cybrid-corrected fibroblasts was performed as described in [24]. Episomal vectors expressing OCT4, SOX2, KLF4, L-MYC, LIN28 and shRNA against p53 were gifts from Shinya Yamanaka (Addgene #27077, 27078, 27080). In vitro differentiation of iPSCs was performed by embryoid bodies and characterised as described in [12]. In vivo teratoma assay was performed by transplanting iPSCs into a vascularized tissue engineering chamber in immuno-deficient rats, as described in [24]. Histological analysis was performed on the teratoma samples after 4 weeks. Copy number variation (CNV) analysis of original fibroblasts and iPSCs was performed using Illumina HumanCore Beadchip arrays. CNV analyses were performed using PennCNV with default parameter settings [25]. Chromosomal aberrations were deemed to involve at least 10 contiguous single nucleotide polymorphisms (SNPs) or a genomic region spanning at least 1MB [25,26].

RGC differentiation
iPSCs were directed for retinal differentiation and enriched for RGCs using MACS THY1.1 microbeads (Miltenyi Biotech) as described in our previous study [14]. On day 30, RGC differentiation efficiency is determined as described previously [14].

Mitochondrial superoxide measurement
Quantification of mitochondrial superoxide was performed on day 30 iPSC-derived RGCs using MitoSOX (Invitrogen) according to the manufacturer's instructions. Briefly, trypsinized cells were stained with Mitosox for 10 minutes at 37⁰C. Subsequently, samples were processed by flow cytometry using the MACSquant (Miltenyi). The median fluorescence intensity was determined using the MACSQuantify software and normalized to unstained control. For each clone, a minimum of 83,000 cells were processed by FACS.

Apoptosis assay
TUNEL assay was performed on day 37-41 iPSCderived RGCs using the In situ Cell death detection kit (Roche) following manufacturer's instructions. Floating apoptotic bodies were collected and the enriched RGCs were harvested by trypsinization. Quantitation of TUNEL positive cells was measured by flow cytometry using the MACSquant (Miltenyi). Apoptosis was assessed in basal conditions, without addition of any stressor or inducer of apoptosis. Generally, a minimum of 10,000 cells per clone were processed by FACS. In clones where <10,000 cells were obtained, experimental repeats were performed to obtain >10,000 cells.

Statistical analysis
All statistical analyses and graphical data were generated using Graphpad Prism software (v5.04, AGING www.graphpad.com) or in the R statistical environment (v3.1.2, https://cran.r-project.org/). Statistical methods utilised were One-way ANOVA followed Dunnett's multiple comparisons test for multiple comparison. Statistical significance was established as * p<0.05 and **p<0.01.

AUTHOR CONTRIBUTIONS
RCBW: concept and design, financial support, collection and/or assembly of data, data analysis and interpretation, manuscript writing, final approval of manuscript. SJ, SK, SYL, SSCH, NJvB, MD, EDS, HHL, LK, LC, DM: collection and/or assembly of data, data analysis and interpretation, final approval of manuscript. AWH, IAT: concept and design, financial support, data analysis and interpretation, final approval of manuscript. AP: concept and design, financial support, data analysis and interpretation, manuscript writing, final approval of manuscript.