Transcriptome and whole genome sequencing proles in Leber’s Hereditary Optic Neuropathy 14484T>C mutation carrying monozygotic twins reveal that prostanoid receptor is a possible modier for LHON manifestation

Leber’s inherited optic neuropathy (LHON) is well known for incomplete penetrance. A pair of monozygotic twins carrying 14484T > C LHON mutation: one displayed LHON characteristics (affected LHON) and the other twin was an unaffected LHON carrier, were studied to identify possible modier(s) for LHON manifestation. Primary broblasts from affected and unaffected monozygotic twins with 14484T > C LHON mutation were treated with different insults to differentiate cellular phenotype between the two broblasts. RNA sequencing of the broblasts indicated differentially expressed genes and whole genome sequencing was used to identify candidates for disease modier. Our results suggested that broblast from unaffected carrier was able to adapt to galactose and hydrogen peroxide insult, while affected broblasts were not. We found reduced expression of total SOD2 with high proportion of inactive SOD2 (acetylated SOD2) in affected LHON broblast, while decreased expression of SIRT3 was detected in affected LHON broblasts treated with combination of insults. Differential expression indicated enrichment of a pathway relating to negative regulator of cell death pathway in unaffected carrier broblast. Expression of receptor for prostaglandin E receptor (PTGER3) was found to be affected by two SNPs. Unaffected LHON broblast possessed rs75523942 that indicates a positive effect on PTGER3 expression, while affected LHON broblast possessed rs496483 that indicates a negative effect on PTGER3 expression. Discordant SNPs on prostaglandin E receptor 3 (PTGER3) were identied as eQTL. This study indicates that prostanoid receptor could be a possible modier for LHON of these twins.


Introduction
Leber's hereditary optic neuropathy (LHON), an inherited mitochondrial disease, typically manifests as painless central visual loss (central scotoma). The visual loss can occur in just one eye or both.
Mitochondrial mutations result in death of retinal ganglionic cells (RGCs). LHON is the most common mitochondrial diseases with an estimated prevalence from 1:30,000 to 1:54,000 worldwide. 1-3 LHON is characterized by its incomplete penetrance with gender bias, affecting about 50% of male mutation carriers and about 10% of female carriers. 4 Nearly 95% of LHON cases are caused by one of the three primary mitochondrial DNA (mtDNA) mutations: 3460G > A in MT-ND1, 11778G > A in MT-ND4 or 14484T > C in MT-ND6 subunit of complex I [5][6][7] . Recovery of the eyesight in LHON is minimal and clinical severity may depend on type of mutation. Prognosis of patients with 11778G > A mutation is the worst while 3460G > A and 14484T > C patients display a milder phenotype 8, 9 .
In earlier studies, a mild to moderate reduction in complex I activity, and complex I dependent ATP synthesis was reported in LHON cybrids carrying either of the above three primary mutations [10][11][12][13] . However, LHON cybrids displayed severe ATP depletion and underwent apoptosis in galactose medium indicating bioenergetics insu ciency under OXPHOS stress condition 14,15 . Antioxidant defenses such as glutathione (GSH) content, activities of glutathione reductase (GR), glutathione peroxidase (GPX) or mitochondrial, superoxide dismutase 2 (SOD2) were reduced in LHON cybrids when growing in galactose medium or exposed to hydrogen peroxide (H 2 O 2 ). These evidences show that cells carrying LHON mutation are vulnerable to oxidative and metabolic stress condition [14][15][16] .
Therefore, increased reactive oxygen species (ROS) production and insu cient ATP production due to complex I defect have been regarded as major contributing factors to RGCs death in LHON. Like other neurons, RGCs demand the highest energy supply for ring action potential along the axon, especially in unmyelinated part 17 . Competent bioenergetics, along with functional antioxidant system are important to supply adequate ATP as well as prevent oxidative damage to mitochondrial DNAs, lipids and proteins under such conditions. Compensatory increase in mitochondrial biogenesis of unaffected LHON carriers from European populations was proposed as an underlying factor for incomplete penetrance in 3460G > A and 11778G > A 18, 19 . Additionally, mtDNA haplogroup backgrounds, 20,21 X-linked loci, 22,23 and potential nuclear modi er gene(s) explored in genome wide studies [24][25][26] were reported to in uence LHON penetrance though ndings varied across populations probably owing to diverse nuclear background in different ethnicities. In addition, environmental factors, such as smoking 27 and exposure to certain drugs 28-30 may also trigger LHON expression.
With advances in high-throughput sequencing technologies, comparison of affected LHON patients and unaffected carriers using omics-wide approaches are of great help to identify disease modifying variants 31 . Monozygotic twins supposedly have identical genetic pro les but discrepancy between disease phenotypes may stem from small errors of DNA replication or variation in epigenetic predisposition 32,33 . Consequences of genetic or epigenetic variations could eventually re ect in gene expression, thus comparing transcriptomics of such discordant monozygotic twins will provide invaluable clues to exploration of disease modifying gene(s), associated signaling pathways and regulatory biological processes. We exploited an oxidative and metabolic insult culture in order to demonstrate a difference in cellular phenotype in 14484T > C mutation carrying monozygotic twins, one unaffected and one affected by LHON. We then created transcriptomic pro les using next-generation sequencing (NGS) based RNA sequencing (RNA-Seq) on the Illumina platform. Differentially expressed genes were identi ed and explored to nd potentially interplaying signaling pathways and biological processes, which might be bene cial in understanding LHON pathogenesis and application for therapeutic interventions.

Results
Effect of hydrogen peroxide or galactose medium treatment on wild type and 14484T > C mitochondrial mutation carrying broblasts Primary dermal broblasts from normal, unaffected LHON and affected LHON individuals were used to determine cellular phenotype. Treatment of hydrogen peroxide in affected LHON broblasts lead to decreased survival compared to normal broblasts ( Fig. 1A and B). Increased production of reactive oxygen species (ROS) with decreased ATP production were found in H 2 O 2 treated affected LHON broblasts compared to unaffected LHON broblasts ( Fig. 1C and F). We subsequently treated all three broblasts with galactose medium; however, no differences in cell survival and ROS production were found ( Fig. 1B and D). However, both unaffected and affected LHON broblasts signi cantly produced less ATP compared to normal broblast when cultured in galactose medium (Fig. 1E). Interestingly, combination of hydrogen peroxide and galactose medium treatment on affected LHON broblasts led to remarkably reduced survival compared to unaffected LHON and normal broblasts, and this nding was consistent with signi cant increase in ROS production compared to either unaffected LHON or wild type broblasts ( Fig. 1B and D).
Combination of hydrogen peroxide and galactose medium treatment on affected LHON broblasts led to reduced Sirtuin 3 (SIRT3) expression and its functionality on superoxide dismutase 2 (SOD2) Lower expression of SOD2 in affected LHON broblast was noted among culture of three broblasts with basal cultured condition ( Fig. 2A and B). Treatment with either H 2 O 2 or galactose medium on affected LHON broblasts led to reduced SOD2 expression in affected group ( Fig. 2A and B). Combination of hydrogen peroxide and galactose medium treatment on affected LHON broblasts further reduced the expression of SOD2 expression compared to either hydrogen peroxide or galactose medium insult; however, a substantial surge of an inactive SOD2 form (lysine acetylation on the 68th amino acid of SOD2) was observed (Fig. 2C). As this Lysine acetylation of SOD2 is known to be regulated by Sirtuin 3 enzyme (SIRT3), we investigated the expression of SIRT3 and found that SIRT3 was marginally reduced in affected LHON broblasts treated with combination of the two insults ( Fig. 3A and B).

RNA-Seq of monozygotic twins with 14484T > C LHON mutation
Total RNA extracted from unaffected and affected LHON broblasts with four different conditions, i.e. untreated, H 2 O 2 , galactose and a combination of both, were included in the RNAseq study. There were 40,112 transcripts detected from differential expression analysis, as shown in Fig. 4A  The 146 discordant SNPs were investigated against expression quantitative trait loci database, but only four SNPs were present in the database with normalised effect size information (Table 1). Among these four SNPs, three were present in unaffected LHON while one SNP was present in affected LHON broblast (Table 2). However, the effect size of only two SNPs, i.e. rs75523942 in unaffected LHON and rs496483 in affected LHON, were in keeping with expression data from RNAseq, and those two were located on prostaglandin E receptor 3 (PTGER3).

Discussion
Culturing cells with galactose medium forces cells to rely on solely OXPHOS bioenergetics, thereby allowing us to exaggerate cellular phenotypes in diseases with mitochondrial dysfunction 34 . We found that cell viability of 14484T > C mitochondrial mutation carrying LHON affected patient broblast was signi cantly lower than that of unaffected carrier in galactose culture supplemented with hydrogen peroxide. This combination of metabolic insult induced signi cantly higher level of ROS in affected LHON. This nding was in keeping with the previous report of mitochondrial mutations induced cybrids cells, in which viability of LHON cybrids were signi cantly reduced as the cells underwent apoptosis in galactose culture after 48 hours of incubation 15 . Increased cellular ROS level along with reduction in cell viability of affected patient broblasts in galactose culture under oxidative stress prompted us to determine functionality of the antioxidant system.
We looked into protein expression of the antioxidant enzyme mitochondrial Manganese superoxide dismutase 2 (SOD2) and also lysine (K)-68 acetylated SOD2, an inactive dismutase form. This K68-AcSOD2 was also previously reported to sensitise cells to peroxide stress 35 . Our results indicated that LHON affected broblasts expressed lower level of total SOD2 and that either peroxide, galactose medium or a combination aggravated reduced level of SOD2 expression, while combination of metabolic insults promoted expression of K68-AcSOD2 form, exclusively in affected LHON broblast. These ndings suggested that aberrant regulation of acetylation switch on SOD2 may occur. were not associated with eQTL in the GTEx database (https://gtexportal.org) 41 .
Differential expression between unaffected and affected LHON broblasts yielded an up and down regulated group of genes enriched on cell death and negative regulator of apoptosis. We subsequently utilised the top twenty of the differentially expressed gene transcript list to search for discordant SNPs as a mean to identify a potential cause of disease phenotype.
Our whole genome sequencing data from twins indicates that discordant SNPs located on CLIC6 and PTGER3 were associated with its expression in GTEx database. CLIC6 is a non-classical ion channel that is structurally homologous to the Glutahione S-transferases Superfamily 42 , and was reported to be highly expressed in human retina 43

Conclusions
In conclusion, we utilised the preconditioned stress culture to differentiate cellular phenotype of patient with mitochondrial disease from the carrier. Discrepancy of the disease phenotype possibly stem from antioxidant mechanism. Cellular capability of the cells to replete pool of NAD + may also contribute to this cellular phenotype.

Sample collection
Dermal broblasts were cultured from a pair of monozygotic twins carrying homoplasmic 14484T > C mutation, and one control individual. One of these twins is an affected LHON patient while the other one does not have a clinical phenotype. The twins were examined by a neuro-ophthalmologist and the control were an independent individual who visited Siriraj Hospital, Thailand for reasons unrelated to eye disease or any other chronic disease conditions. Determination of cellular ROS content Cellular ROS content was determined using hydroethidine (aka. dihydroethidium or DHE) assay (Invitrogen) according to manufacturer's protocol. Brie y, broblasts were seeded and treated with different culture condition in the same fashion as the treatment protocol for cell viability assay.
Fibroblasts were subsequently stained with 10µM of DHE for 30 min at 37°C in the dark. DHE oxidation was determined by Synergy H1 Hybrid Multiplate Reader (BioTek) at the excitation/emission of 518nm/606nm. Cellular ROS content is expressed as percentage of the DHE oxidation of untreated controls in glucose or galactose medium respectively.

Determination of whole cell ATP content
Whole cell ATP content was determined by using ATP bioluminescence assay kit (Sigma) according to manufacturer's protocol. Fibroblasts were seeded in precondition of either high glucose DMEM or 25mM galactose media for 48hours. The medium was replaced by fresh medium supplemented with/without H 2 O 2 treatment for 90 minutes. Fibroblasts were washed in cold PBS for twice, ATP was extracted in 0.1% TritonX-100 ATP extraction buffer, as described by Koszegi et al. 49 . Supernatant were kept on ice, stored at -20°C and used for quanti cation of ATP and protein content. Protein content was determined by Bradford assay (Bio-Rad). Whole cell ATP level was expressed by µmol/mg protein and percentage of ATP content in galactose was expressed by ratio of whole cell ATP level in galactose medium to that of glucose medium. Percentage of ATP content under oxidative stress was expressed by ratio of whole cell ATP level in H 2 O 2 exposed cells to that of untreated cells.

Whole cell protein extraction and western blotting
Fibroblasts cultured in their respective experimental conditions were trypsinised and washed in PBS. Cell pellets were dissociated in RIPA lysis buffer with freshly added protease inhibitors (Sigma) and 1mM of phenylmethylsulfonyl uoride (PMSF). Extracted protein concentration were measured using Bradford's method and stored at -80°C until use. Whole cell protein extracts were resolved in 10% SDS-PAGE. The PVDF membrane was blocked in 5% skimmed milk in 0.1% Tween-PBS (PBST) buffer for 1 hour to reduce non-speci c binding and the blotted membranes were incubated with respective primary antibodies at 4°C overnight. The membranes were subsequently incubated with horseradish peroxidase (HRP)conjugated secondary antibody Chemiluminescence HRP solution (Millipore) was applied to visualize target protein bands on the Image Quant LAS 4000 (GE healthcare). Band intensities of proteins were determined using Image J software (NCBI). Primary and secondary antibodies used were SIRT3 (Cell Signaling Technologies); β-actin (Sigma); SOD2, Lys-68 acetylated SOD2, goat anti-rabbit and anti-mouse IgG HRP antibodies (Abcam).

RNA extraction and transcriptome pro ling
Total RNA was extracted from broblasts grown in basal glucose culture, and culture with hydrogen peroxide, galactose or combination using RNeasy Mini Kit (Qiagen) according to manufacturer's protocol. RNA Sequencing (RNA-Seq) was performed by Illumina using next generation sequencing. FASTQ les were uploaded to a web-based RNA-seq analysis pipeline by Basepair (https://www.basepairtech.com/).
Short-read data were aligned to the reference genome from UCSC (GRCh37) using STAR aligner 50 and differential counts obtained from HT-seq count were determined by DESeq2 51 . List of differentially expressed genes from twins with different conditions were pre-ranked based on p-value and log-2 fold change prior to Gene set enrichment analysis (GSEA Availability of data and materials The datasets used and/or analysed during the current study are available from the corresponding author on reasonable request.

Competing interests
Authors have declared no con ict of interest. Authors' contributions KZT did experiment and data analysis. JT did experiment and data analysis. SK did characterize mitochondrial DNA mutation. HG did whole genome sequencing and transcriptomic data analysis. SR did transcriptomic analysis. WS critically read manuscript. WC diagnosed LHON patients, collected dermal broblasts. PL critically read manuscript. BS planned experiment, analysed data and wrote manuscript. CM planned experiment, analysed data and wrote manuscript.  Western blot analysis of SOD2, acetylated SOD2 at the 68th amino acid of lysine (AcK68-SOD2) and βactin (A) on protein extract from normal, unaffected LHON and affected LHON broblasts, which were treated with galactose, hydrogen peroxide or combination of both metabolic insults. Densitometric analysis of the protein band of SOD2 that was normalized against β-actin (B) and AcK68-SOD2 that was normalized against total SOD2 (C).

Figure 3
Western blot analysis of SIRT3 and β-actin (A) on protein extract from normal, unaffected LHON and affected LHON broblasts, which were treated with galactose, hydrogen peroxide or combination of both metabolic insults. Densitometric analysis of the protein band of SIRT3 (B) that was normalized against βactin.