Abstract

Leber's hereditary optic neuropathy (LHON) is thought to be the most common disease resulting from mitochondrial DNA (mtDNA) point mutations, and transmitochondrial cytoplasmic hybrid (cybrid) cell lines are the most frequently used model for understanding the pathogenesis of mitochondrial disorders. We have used oligonucleotide microarrays and a novel study design based on shared transcripts to allocate transcriptomal changes into rho-zero-dependent, cybridization-dependent and LHON-dependent categories in these cells. The analysis indicates that the rho-zero process has the largest transcriptomal impact, followed by the cybridization process, and finally the LHON mutations. The transcriptomal impacts of the rho-zero and cybridization processes preferentially and significantly affect the mitochondrial compartment, causing upregulation of many transcripts involved in oxidative phosphorylation, presumably in response to the mtDNA depletion that occurs at the rho-zero step. Nine LHON-specific transcriptional alterations were shared among osteosarcoma cybrids and lymphoblasts bearing LHON mutations. Notably, the aldose reductase transcript was overexpressed in LHON cybrids and lymphoblasts. Aldose reductase is also overexpressed in diabetic retinopathy, leading to optic nerve and retinal complications. The LHON-specific increase in transcript level was confirmed by quantitative reverse transcription–polymerase chain reaction (RT–PCR), and a western blot confirmed a higher level of aldose reductase in mutant mitochondria. One product of aldose reductase is sorbitol, which has been linked to osmotic stress, oxidative stress and optic neuropathy, and sorbitol levels were increased in LHON cybrids. If these results are confirmed in patient tissues, aldose reductase inhibitors could have some therapeutic value for LHON.

Introduction

Leber's hereditary optic neuropathy (LHON) is a maternally inherited disorder characterized by a primary degeneration of the retinal ganglion cells (RGCs) and atrophy of the optic nerve (Carelli et al., 2004). Central vision is mostly affected, due to a preferential death of the small nerve fibres of the papillomacular bundle (Sadun et al., 2000). LHON is most commonly caused by three pathogenic mitochondrial DNA (mtDNA) point mutations (11778, 3460 and 14484) which alter genes coding for complex I subunits in the mitochondrial electron transport chain (Wallace et al., 1988; Howell et al., 1991; Huoponen et al., 1991; Johns et al., 1992). While the primary genetic cause of this devastating disease has been known for some time (Wallace et al., 1988), the precise mechanism by which these mutations, through an alteration of complex I, result in blindness essentially remains an enigma.

Human transmitochondrial cytoplasmic hybrids (cybrids), first described by King and Attardi (1989) using the osteosarcoma-derived 143B.TK cells (hereafter unfused controls), have been a valuable model for the study of mitochondrial pathophysiology. The appeal of the cybrid cell model relates to the possibility of dissecting the donor mtDNA from the original nuclear background, and thus studying the mitochondrial mutation-dependent differences in isolation, i.e. in cells with a common presumably constant and ‘neutral’ nuclear background. However, neither a quantitative and systematic investigation of the transcriptomal effects of the cybridization process nor LHON mutations have been presented to date.

In principle, it seems reasonable that the cybridization process could cause substantial cellular stress and produce major effects on gene expression, due to the drastic procedure of endogenous mtDNA depletion by long-term exposure to the mutagen ethidium bromide (EtBr), in order to obtain rho-zero cells from the parental cell line. The cybridization process also entails the enucleation of the donor cells by cytochalasin B treatment, and the subsequent fusion of the rho-zero recipient cells with the cytoplasts obtained after donor cells enucleation, which may cause damage and disorganization to multiple cellular organelles and membranes. Also, the parental cell line nucleus, i.e. 143B.TK rho-zero 206, is the nucleus of an aneuploid osteosarcoma cell, and its inherent genetic instability could produce variability in gene expression.

Despite these concerns, and possibly because of the paucity of transmitochondrial animal models currently available, cybrids remain a workhorse to investigate human mitochondrial pathophysiology (Hayashi et al., 1991; King et al., 1992; Mariotti et al., 1994; Hofhaus et al., 1996; Swerdlow et al., 1997; Wong and Cortopassi, 1997; Barrientos and Moraes, 1999; Williams et al., 1999; Brown et al., 2000; Gajewski et al., 2003). Cybrids repopulated with normal mtDNA appear to function normally, having no decrease in growth rate, oxygen consumption or activities of respiratory enzymes when compared with the parental 143B.TK cells (King and Attardi, 1989; Carelli et al., 2002b).

Microarray analysis is a powerful technique that can enable researchers to examine the expression of thousands of transcripts in parallel. Because cybrids share a common nucleus, and thus in principle no genetic mutations between lines, these would seem to be the ideal system in which to isolate the mtDNA mutation-dependent effects on nuclear gene expression. We have utilized microarray analysis to examine the altered expression of nuclear genes caused by interaction with mtDNA mutations that cause LHON. In the process of the study, we discovered a very large background effect, that through our analytical tests was attributable to the rho-zero and cybridization processes. This same study design and analysis also allowed the isolation of LHON-dependent transcriptomal changes independent of the rho-zero and cybridization effects.

Materials and methods

Cell lines and culture conditions

The parental osteosarcoma 143B.TK (mtDNA haplogroup X) and the 143B206 rho-zero cell lines from which cybrids were derived were a kind gifts from G. Attardi (Cal Tech, Pasadena, CA) and M. Zeviani (National Neurological Institute ‘Carlo Besta’, Milan, Italy), respectively. Cybrid control cell lines were constructed as described elsewhere (King and Attardi, 1989) by fusing 143B206 rho-zero cells with enucleated fibroblasts bearing mtDNA with no disease-related mutations. The three control cybrid cell lines are derived from three different individuals (bearing no disease-related mtDNA mutations) designated HPC7 (mtDNA haplogroup H), H1959 (mtDNA haplogroup T) and HGA13 (mtDNA haplogroup J) as previously reported (Carelli et al., 2002b). Five LHON cybrid cell lines were constructed from five individual patients; HCT1, HPE9 and HFF3 harboured the 11778 pathogenic mutation (the first two being J and the last being of the U haplogroups, respectively) and HMM5 and RJ206 contained the 3460 pathogenic mutation (H and T haplogroups, respectively). Clones were obtained after growth in selective medium (lacking uridine and with bromodeoxyuridine) and the clones obtained underwent mtDNA investigation to confirm the source of the mtDNA as described previously (Torroni et al., 1997). All osteosarcoma cell lines for experiments were maintained in Dulbecco's modified Eagle's medium (Invitrogen, Carlsbad, CA) supplemented with 10% fetal bovine serum (FBS), 1 mM sodium pyruvate, 50 µg/ml uridine and 40 µg/ml gentamicin. All cybrid cell lines were generated >5 years ago, cells grew normally and were mycoplasma free. Cell lines were stored in liquid nitrogen until needed for experiments, at which time they were revived, and grown for a limited number of passages (∼10) to isolate RNA for microarray analysis. Three control lymphoblast cell lines were obtained from Coriell Cell Repositories (Camden, NJ) and four lymphoblast cell lines harbouring LHON mtDNA mutations from affected individuals were utilized. Cell lines 910615 and 11605 (the latter from Coriell) contained the 3460 mtDNA mutation; cell lines 980002 and 980004 contained the 11778 mtDNA mutation (Wong et al., 2002). Lymphoblast cells were maintained in RPMI 1640 medium (Invitrogen) supplemented with 15% FBS, 1 mM sodium pyruvate, 50 µg/ml uridine, 4 mM glutamine, and 5 µM penicillin and streptomycin.

cRNA labelling and microarray

We utilized Affymetrix U95Av2 oligonucleotide chips which represent 12 599 human transcipts (9091 unique human genes) to identify differentially expressed genes. Microarray samples were made as per the Affymetrix protocol. Briefly, total RNA was isolated from ∼8×106 cells using a Qiagen RNeasy mini kit (Valencia, CA); 10 µg of total RNA was then used to generate double-stranded cDNA using an oligo(dT24) primer containing a T7 polymerase recognition site. The cDNA was then used as a template for an in vitro transcription reaction (ENZO Farmingdale, NY) using biotinylated CTP and UTP to create labelled cRNA. The cRNA (20 µg) is then fragmented and hybridized to Affymetrix U95Av2 Gene Chips, which are then washed and stained with avidin-labelled phycoerytherin. The chips are then scanned with an Affymetrix GeneChip scanner and the raw data are processed by Microarray Suite 4.0 (Affymetrix).

Data analysis

Microarray data files were analysed using DNA-Chip Analyzer (dChip) software (Li and Wong, 2001). Differentially expressed genes were identified by comparing GeneChips designated as baseline (e.g. unfused control cell lines) with GeneChips that represent the experimental parameter (e.g. control cybrid or rho-zero cell lines) using the following criteria, a difference in mean fluorescence intensity between baseline and experimental ≥30, and a P-value <0.05. For each GeneChip, biotin-labelled cRNA from one cell line was hybridized according to the manufacturers' (Affymetrix) specifications. Six GeneChips were hybridized to samples from the unfused control 143B.TK osteosarcoma line, five GeneChips were hybridized to samples from three fused control cell lines [HGA13 (three GeneChips), HPC7 (one GeneChip) and H1959 (one GeneChip)], two GeneChips were hybridized to samples from the 143B.TK rho-zero 206 cell line, and 13 GeneChips were hybridized to samples obtained from five fused control lines bearing a mtDNA mutation which causes LHON [HFF3 (three GeneChips), HPE9 (one GeneChip), HCT1 (three GeneChips), HMM5 (three GeneChips) and RJ206 (three GeneChips)]. Three GeneChips were used for each lymphoblast control cell line and one GeneChip was used for each lymphoblast cell line bearing LHON mutations. Gene profile analysis was then done to determine if there are any relationships between the genes with altered expression based on their subcellular localization or biological process using the program Onto-Express (Khatri et al., 2002).

Quantitative reverse transcriptase—polymerase chain reaction (RT–PCR)

Total RNA was prepared from 8 × 106 cells using an RNeasy minikit (Qiagen). Superscript reverse transcriptase II (Invitrogen Carlsbad, CA) was used to create cDNA starting with 1 µg of total RNA. Aldose reductase and β-actin were amplified using the following primers: aldose reductase forward 5′-TGAACTGAGCAGCCAGGATA-3′ and reverse 5′-AGGGCTCTTGAGATCCAACAT-3′, β-actin forward 5′-ACGGCATCGTCACCAA CTGG-3′ and reverse 5′-TTCATGAGGTAGTCAGTCAGG-3′. PCR was carried out using a Roche Lightcycler (Indianapolis, IN) annealing at 58°C for 5 s, and extension at 72°C for 15 s for 40 cycles. Samples were quantified by generating a standard curve using cDNA transcribed from brain RNA (Ambion, Austin, TX).

Western blot analysis

Mitochondrial fractions were obtained from osteosarcoma cybrids according to the procedure of Trounce et al. (1996). Lysates were prepared and then analysed by western blot as previously described (Danielson et al., 2002). Briefly, 20 µg of mitochondrial lysate was loaded onto a 10% pre-cast Ready Gel (Bio-Rad, Hercules, CA) for SDS–PAGE, and afterwards the sample was transferred to a PVDF membrane. The blot was blocked with 5% milk and then probed with an aldose reductase-specific antibody (a kind gift from J. D. Hayes, University of Dundee) at a 1 : 2000 dilution. The blot was developed by enhanced chemiluminescence according to the manufacturer's specifications (Amersham, Buckinghamshire, UK). The average band density was measured using Quantity One software from Bio-Rad. The density values for each band were normalized to the positive control on each blot (pure aldose reductase, Wako Chemicals). In order to check for equal levels of protein for all samples, the blot probed for aldose reductase was stripped using a stripping buffer (Pierce, Rockford, IL) according to the manufacturer's specifications, and the blot was then reprobed as described above using a cytochrome c-specific antibody (BD Pharmingen, San Diego, CA).

Sorbitol measurement

Sorbitol levels were measured as previously described in Bergmeyer et al. (1974). All chemicals and enzymes were purchased from Sigma (St Loius, MO): briefly 5 × 106 osteosarcoma cells were harvested and then washed three times with phosphate-buffered saline (PBS). The cell pellet was suspended in 100 µl of water and the sample was deproteinated through addition of 100 µl of ice-cold 1 M perchloric acid. The solution was mixed and centrifuged at 3000 g for 15 min. A 100 µl aliquot of 2 M potassium hydroxide was added to the supernatant and the solution was then centrifuged at 7000 g for 10 min to remove any precipitate. A 300 µl aliquot of sample was combined with 667 µl of buffer (0.1 M sodium pyrophosphate pH 9.5), 33 µl of NAD (5 mg/ml) and 17 µl of sorbitol dehydrogenase (SDH) (30 mg/ml). The absorbance at 340 nm was measured before addition of the SDH and 25 min after addition, when the reaction had consumed all substrates. The difference in absorbance of the sample (before and after addition of the SDH) at 340 nm was then used to calculate the concentration of sorbitol. A standard curve of sorbitol concentration was constructed using solutions containing 100, 50, 10, 1 and 0.1 µg/ml of sorbitol.

Results

Attribution of transcriptomal variability to rho-zero-, cybrid- and LHON-dependent steps through study design

In order to isolate the transcriptomal effects of the LHON mutations from the rho-zero and cybridization processes, we considered the steps involved in the process (Fig. 1). Thus, we organized our osteosarcoma microarray comparisons into four sets: the unfused controls (not rho-zero, not cybridized, no LHON mutations); the rho-zero cell line (rho-zero derived, not cybridized, no LHON mutations); the control cybrids (rho-zero derived, cybridized, no LHON mutations); and the LHON cybrids (rho-zero derived, cybridized, with LHON mutations). We then used the principle that to assign transcriptomal changes to a specific process (e.g. cybridization or LHON), those transcriptomal changes had to be statistically significant and shared in two different groups of cells that had independently undergone that process.

Fig. 1

Microarray comparisons done using osteosarcoma cell lines. Schematic diagram representing the microarray comparisons done, the number of significantly altered genes for each comparison and the presumed reason for the altered expression. Four osteosarcoma cell types were analysed by microarray analysis of unfused controls, rho-zero, control cybrids and LHON cybrids; each cell type is represented by a circle with a smaller circle inside to represent the mitochondrial genome (mtDNA), thus the rho-zero cell line contains a broken circle to represent that it contains no mtDNA, and the mtDNA in the LHON cybrid cell line contains a star to represent the LHON point mutations. Letters are used to designate the individual microarray comparisons that were done (described in ‘Comparisons’ and ‘Combined comparisons’). The cell lines used are listed as well as the number of genes with significantly altered expression (P < 0.05). Any treatments or mutations which may be the cause of the altered gene expression are also listed for each comparison. The results of individual microarray comparions were compared (‘Combined comparisons’ table) so as to confirm and isolate the specific effects on gene expression caused by the treatments to create rho-zero and cybrid cell lines as well as the effects due to LHON mutations. ∩ designates that microarray experiments were compared to identify genes that were shared between two or more microarray experiments; the expression of these genes was significantly altered in the same direction for both experiments. The numbers of genes that were significantly altered in both comparions are listed as well as the effect that is attributed to the altered gene expression.

Fig. 1

Microarray comparisons done using osteosarcoma cell lines. Schematic diagram representing the microarray comparisons done, the number of significantly altered genes for each comparison and the presumed reason for the altered expression. Four osteosarcoma cell types were analysed by microarray analysis of unfused controls, rho-zero, control cybrids and LHON cybrids; each cell type is represented by a circle with a smaller circle inside to represent the mitochondrial genome (mtDNA), thus the rho-zero cell line contains a broken circle to represent that it contains no mtDNA, and the mtDNA in the LHON cybrid cell line contains a star to represent the LHON point mutations. Letters are used to designate the individual microarray comparisons that were done (described in ‘Comparisons’ and ‘Combined comparisons’). The cell lines used are listed as well as the number of genes with significantly altered expression (P < 0.05). Any treatments or mutations which may be the cause of the altered gene expression are also listed for each comparison. The results of individual microarray comparions were compared (‘Combined comparisons’ table) so as to confirm and isolate the specific effects on gene expression caused by the treatments to create rho-zero and cybrid cell lines as well as the effects due to LHON mutations. ∩ designates that microarray experiments were compared to identify genes that were shared between two or more microarray experiments; the expression of these genes was significantly altered in the same direction for both experiments. The numbers of genes that were significantly altered in both comparions are listed as well as the effect that is attributed to the altered gene expression.

Isolation of rho-zero- and cybridization-dependent genes

Altered gene expression in rho-zero cells was determined by comparing the unfused control cell line with the rho-zero cell line (Fig. 1 comparison A). A total of 1843 transcripts were significantly altered in rho-zero cells versus unfused control cells (see Supplementary material at Brain Online). We also performed microarray analysis on two types of osteosarcoma cybrids; control cybrids made with mitochondria from healthy control individuals, and LHON cybrids made from mitochondria from patients with LHON. Thus, cybridization-dependent transcriptional changes were defined as those significant changes shared and thus confirmed (i.e. in the same transcript and in the same direction) between unfused controls versus control cybrids (Fig. 1 comparison B), and also between unfused controls and LHON cybrids (Fig. 1 comparison C) represented by B ∩ C in Fig. 1. We observed 1934 statistically significant changes at the P < 0.05 level in comparison B, and 1042 in comparison C (Fig. 1). Of these transcripts, 689 transcripts were significantly altered in the same direction in both of these comparisons. However, the expression of 26 of these genes was ‘ambiguous’ in that they were significantly altered in the comparisons to identify both cybrid (B ∩ C) as well as LHON-dependent effects (C ∩ D discussed in the section on LHON-dependent effects), i.e. altered by both cybridization and LHON. We assigned 22 of these changes as cybrid dependent, as they were not shared with LHON-dependent changes in lymphoblasts (discussed below), which have not undergone a cybridization process. The remaining four genes were classified as LHON dependent as their expression was altered in the same direction in lymphoblast cell lines bearing LHON mutations. Thus the expression of 685 genes (689 – 4 LHON-dependent genes) was consistently altered by the cybridization process (see Supplementary material), and these were designated as cybridization dependent.

The rho-zero process produces the majority of the cybridization-dependent transcriptomal signature

We then inferred what fraction of the cybridization-dependent alterations were the result of the rho-zero process, by investigating the transcriptional changes shared between the rho-zero process (the A comparison) and the cybridization process (the B ∩ C comparison), i.e. A ∩ (B ∩ C). Sixty-two percent of the cybridization-dependent transcriptional changes, 422 of the 685, were shared between the cybridization process and the rho-zero process. Since cybrids are derived from rho-zero cells, this demonstrates that the rho-zero process has a huge transcriptomal signature, that dominates the transcriptomal signature of the cybridization process itself. It also demonstrates that the rho-zero process causes stable transcriptional alterations that are preserved even past the repopulation step of cybridization in the control and mutant cybrids.

Cybridization preferentially and significantly affects genes in the mitochondrial compartment

We classified the 685 transcripts identified as cybridization dependent using the Onto-Express program (Khatri et al., 2002), to determine whether transcripts were significantly clustered with respect to any particular subcellular compartment or biological process. Cybridization strongly affected transcripts in the mitochondrial subcellular category, four orders of magnitude (P = 0.000001) more significantly than the next most affected compartment, the cytoplasm (P = 0.01). Thus cybridization produces a very strong and significant ‘imprint’ on mitochondrially targeted transcripts, and the effect on mitochondrially targeted transcripts is much stronger than on any other intracellular compartment (Table 1). The 42 cybridization-dependent genes identified by Onto-Express as mitochondrial are listed in Table 2 along with the fold change observed for that gene in cybrid and rho-zero cells. The majority of these transcripts are functionally involved in oxidative phosphorylation and upregulated, including subunits of complexes I–V of the electron transport chain and the tricarboxylic acid cycle (TCA cycle). The transcriptional ‘imprinting’ of the rho-zero process on cybrids is illustrated by the fact that nearly all the genes in Table 2 are significantly altered in expression in both rho-zero and cybrid cells.

Table 1

Subcellular localizations of transcripts affected in cybrid cells, ranked by number of unique genes

Subcellular localization
 
Unique genes*
 
Total**
 
P-value
 
Mitochondrion 42 218 0.000001 
Cytoplasm 40 359 0.01761 
Integral to plasma membrane 27 617 0.02092 
Extracellular matrix 10 75 0.04154 
Inner membrane 17 0.02047 
Subcellular localization
 
Unique genes*
 
Total**
 
P-value
 
Mitochondrion 42 218 0.000001 
Cytoplasm 40 359 0.01761 
Integral to plasma membrane 27 617 0.02092 
Extracellular matrix 10 75 0.04154 
Inner membrane 17 0.02047 
*

Unique genes represents the number of genes with significantly altered expression for a particular category

**

Total represents the total number of genes that comprise that particular category as determined by the Gene Ontology consortium.

Table 2

Significantly altered cybrid-dependent genes localized to the mitochondria

Gene
 
Cybrid FC*
 
Rho-zero FC*
 
Aconitase 2, mitochondrial 1.78 NS 
Adenylate kinase 2 2.05 2.47 
AFG3 (ATPase family gene 3, yeast)-like 2 1.93 2.66 
ATP synthase, mitochondrial F0 complex, subunit c (subunit 9) isoform 3 1.68 1.38 
ATP synthase, mitochondrial F0 complex, subunit d 1.52 1.22 
ATP synthase, mitochondrial F0 complex, subunit f, isoform 2 1.49 1.90 
ATP synthase, mitochondrial F0 complex, subunit g 1.86 1.78 
ATP synthase, mitochondrial F1 complex, O subunit 1.39 1.47 
Carbamoyl-phosphate synthetase 1, mitochondrial −1.76 −1.80 
Cytochrome c 1.40 2.03 
Cytochrome c oxidase subunit VIb 1.42 1.29 
Cytochrome c oxidase subunit VIIa polypeptide 2 (liver) 1.27 1.19 
Cytochrome c oxidase subunit VIIa polypeptide 2-like 1.39 NS 
Cytochrome c oxidase subunit VIIb 1.46 1.38 
Cytochrome c oxidase subunit VIII 1.32 1.33 
DNA segment on chromosome 19 (unique) 1177 expressed sequence 1.49 3.18 
Fibroblast growth factor (acidic) intracellular binding protein 1.45 1.49 
Glutathione reductase 2.49 2.31 
Glycine amidinotransferase −3.31 NS 
Heat shock 10 kDa protein 1 (chaperonin 10) 1.28 1.55 
Isocitrate dehydrogenase 3 (NAD+) α 1.86 1.81 
Mitochondrial ribosomal protein S18B 1.57 1.31 
NADH dehydrogenase (ubiquinone) 1α subcomplex, 6 (14 kDa, B14) 1.68 1.49 
NADH dehydrogenase (ubiquinone) 1β subcomplex, 1 (7 kDa, MNLL) 1.40 1.44 
NADH dehydrogenase (ubiquinone) 1β subcomplex, 3 (12 kDa, B12) 1.97 2.15 
NADH dehydrogenase (ubiquinone) Fe-S protein 3 (30 kDa) 1.52 NS 
NADH dehydrogenase (ubiquinone) Fe-S protein 5 (15kD) 1.45 1.26 
NADH dehydrogenase (ubiquinone) flavoprotein 2 (24 kDa) 1.87 1.71 
Novel RGD-containing protein 2.04 1.46 
Oxidase (cytochrome c) assembly 1-like −1.55 −1.48 
Phosphoenolpyruvate carboxykinase 2 (mitochondrial) −2.27 −1.92 
Pyruvate carboxylase 1.55 NS 
Solute carrier family 25 (mitochondrial carrier; adenine nucleotide translocator), member 5 1.68 1.71 
Solute carrier family 25, member 13 (citrin) 1.61 1.98 
Succinate dehydrogenase complex, subunit A, flavoprotein (Fp) −1.54 −1.24 
Succinate-CoA ligase, ADP-forming, β subunit 2.23 2.38 
Translocase of inner mitochondrial membrane 17 (yeast) homologue A 1.72 2.34 
Translocase of outer mitochondrial membrane 20 (yeast) homologue 1.79 1.77 
Translocase of outer mitochondrial membrane 34 −1.53 1.53 
Ubiquinol-cytochrome c reductase core protein II 1.61 1.36 
Ubiquinol-cytochrome c reductase hinge protein 1.36 1.36 
Voltage-dependent anion channel 3 1.50 1.63 
Gene
 
Cybrid FC*
 
Rho-zero FC*
 
Aconitase 2, mitochondrial 1.78 NS 
Adenylate kinase 2 2.05 2.47 
AFG3 (ATPase family gene 3, yeast)-like 2 1.93 2.66 
ATP synthase, mitochondrial F0 complex, subunit c (subunit 9) isoform 3 1.68 1.38 
ATP synthase, mitochondrial F0 complex, subunit d 1.52 1.22 
ATP synthase, mitochondrial F0 complex, subunit f, isoform 2 1.49 1.90 
ATP synthase, mitochondrial F0 complex, subunit g 1.86 1.78 
ATP synthase, mitochondrial F1 complex, O subunit 1.39 1.47 
Carbamoyl-phosphate synthetase 1, mitochondrial −1.76 −1.80 
Cytochrome c 1.40 2.03 
Cytochrome c oxidase subunit VIb 1.42 1.29 
Cytochrome c oxidase subunit VIIa polypeptide 2 (liver) 1.27 1.19 
Cytochrome c oxidase subunit VIIa polypeptide 2-like 1.39 NS 
Cytochrome c oxidase subunit VIIb 1.46 1.38 
Cytochrome c oxidase subunit VIII 1.32 1.33 
DNA segment on chromosome 19 (unique) 1177 expressed sequence 1.49 3.18 
Fibroblast growth factor (acidic) intracellular binding protein 1.45 1.49 
Glutathione reductase 2.49 2.31 
Glycine amidinotransferase −3.31 NS 
Heat shock 10 kDa protein 1 (chaperonin 10) 1.28 1.55 
Isocitrate dehydrogenase 3 (NAD+) α 1.86 1.81 
Mitochondrial ribosomal protein S18B 1.57 1.31 
NADH dehydrogenase (ubiquinone) 1α subcomplex, 6 (14 kDa, B14) 1.68 1.49 
NADH dehydrogenase (ubiquinone) 1β subcomplex, 1 (7 kDa, MNLL) 1.40 1.44 
NADH dehydrogenase (ubiquinone) 1β subcomplex, 3 (12 kDa, B12) 1.97 2.15 
NADH dehydrogenase (ubiquinone) Fe-S protein 3 (30 kDa) 1.52 NS 
NADH dehydrogenase (ubiquinone) Fe-S protein 5 (15kD) 1.45 1.26 
NADH dehydrogenase (ubiquinone) flavoprotein 2 (24 kDa) 1.87 1.71 
Novel RGD-containing protein 2.04 1.46 
Oxidase (cytochrome c) assembly 1-like −1.55 −1.48 
Phosphoenolpyruvate carboxykinase 2 (mitochondrial) −2.27 −1.92 
Pyruvate carboxylase 1.55 NS 
Solute carrier family 25 (mitochondrial carrier; adenine nucleotide translocator), member 5 1.68 1.71 
Solute carrier family 25, member 13 (citrin) 1.61 1.98 
Succinate dehydrogenase complex, subunit A, flavoprotein (Fp) −1.54 −1.24 
Succinate-CoA ligase, ADP-forming, β subunit 2.23 2.38 
Translocase of inner mitochondrial membrane 17 (yeast) homologue A 1.72 2.34 
Translocase of outer mitochondrial membrane 20 (yeast) homologue 1.79 1.77 
Translocase of outer mitochondrial membrane 34 −1.53 1.53 
Ubiquinol-cytochrome c reductase core protein II 1.61 1.36 
Ubiquinol-cytochrome c reductase hinge protein 1.36 1.36 
Voltage-dependent anion channel 3 1.50 1.63 
*

FC = fold change.

NS = not significant.

Cybridization preferentially affects mitochondrial biochemical pathways

We also used the ‘bioprocess’ function of Onto-Express to sort and rank the biochemical pathways most altered by the 685 cybridization-dependent changes. Genes involved in electron transport, the TCA cycle, energy pathways, ATP biosynthesis and aerobic regulation all gave very strong signals (Table 3), and the vast majority of the genes in these categories were significantly upregulated, by a ratio of 9 : 1, relative to downregulated. The results suggest that the nucleus perceives a decreased mitochondrial function as a result of the rho-zero process (which involves a long, slow mtDNA depletion with the nuclear and mitochondrial mutagen EtBr), and that it compensates by overexpressing mitochondrially targeted and particularly oxidative phosphorylation (OXPHOS)-dependent transcripts, and that this overexpression persists in the cybrid cells.

Table 3

Biological processes significantly altered in cybrid cell lines, ranked by number of unique genes

Biological process
 
Unique genes*
 
Total**
 
P-value
 
Electron transport 17 134 0.01790 
Intracellular protein transport 15 127 0.03845 
Transcription from Pol II promoter 15 126 0.04328 
Regulation of transcription from Pol II promoter 14 110 0.04264 
Negative regulation of cell proliferation 12 86 0.03568 
Anti-apoptosis 39 0.02766 
Tricarboxylic acid cycle 14 0.00011 
Energy pathways 52 0.02749 
Proton transport 26 0.01492 
ATP biosynthesis 12 0.01116 
Regulation of cell growth 27 0.02804 
Aerobic respiration 0.00495 
Biological process
 
Unique genes*
 
Total**
 
P-value
 
Electron transport 17 134 0.01790 
Intracellular protein transport 15 127 0.03845 
Transcription from Pol II promoter 15 126 0.04328 
Regulation of transcription from Pol II promoter 14 110 0.04264 
Negative regulation of cell proliferation 12 86 0.03568 
Anti-apoptosis 39 0.02766 
Tricarboxylic acid cycle 14 0.00011 
Energy pathways 52 0.02749 
Proton transport 26 0.01492 
ATP biosynthesis 12 0.01116 
Regulation of cell growth 27 0.02804 
Aerobic respiration 0.00495 
*

Unique genes represents the number of genes with significantly altered expression for a particular category

**

Total represents the total number of genes that comprise that particular category as determined by the Gene Ontology consortium.

LHON-specific alterations in gene expression in osteosarcoma cybrids

We defined LHON-specific alterations in osteosarcoma cells as those shared between unfused controls versus LHON cybrids (Fig. 1 comparison C) and control cybrids versus LHON cybrids (comparison D), i.e. shared in two independent control versus mutant comparisons. There were 1042 significant transcriptional changes in comparison C, 1306 in comparison D, and 118 shared in the same direction. As previously mentioned, the expression of 26 of these genes was ambiguous in that they were significantly altered in both comparisons to identify cybrid- (B ∩ C) and LHON-dependent (C ∩ D) effects. Four of these ambiguous genes were classified as LHON dependent as they were shared with LHON-dependent changes in lymphoblasts (discussed below) which are not affected by cybridization; the rest (22) were classified as cybrid dependent. Thus the expression of 96 genes (118–22 ambiguous genes classified as cybrid dependent) was specifically altered by LHON mutations, and these were defined as LHON dependent (Supplementary material).

Identification of LHON-dependent alterations in gene expression shared among osteosarcoma and lymphoblasts

Because the rho-zero process produces noise specifically in mitochondrial transcripts, we investigated LHON-dependent changes in lymphoblasts uncomplicated by the rho-zero process. Four LHON lymphoblast cell lines were microarrayed, and compared with three control lymphoblast cell lines. A total of 715 statistically significantly altered transcripts were identified (Supplementary material).

We compared the 96 significantly LHON-dependent transcripts identified from osteosarcoma cybrids with the 715 significantly LHON-dependent transcripts identified from lymphoblasts. Nine transcripts were significantly altered in the same direction in a LHON-dependent manner and shared between osteosarcoma and lymphoblasts (Table 4). Aldose reductase, methylenetetrahydrofolate dehydrogenase (MTHFD), histone 2A, immunoglobulin superfamily member 3, integral membrane protein 2B (ITM2B), lipin 1, TUBA scaffold protein, sialyltransferase 1 and Raf1 were each LHON dependent by this analysis.

Table 4

LHON-specific changes shared between osteosarcoma cybrids and lymphoblasts

Gene
 
Function
 
Lymphoblast FC*
 
Osteosarcoma FC*
 
Aldose reductase Catalyses the reduction of a number of aldehydes 1.62 1.57 
 Associated with retinopathy, neuropathy and microangiopathy in diabetes   
Scaffold protein TUBA Functions to bring together dynamin with actin regulatory proteins −2.22 −1.58 
Methylenetetrahydrofolate dehydrogenase (NADP+ dependent) N10-Formyl tetrahydrofolate biosynthesis −1.38 −1.57 
 Needed for reactions requiring addition of 1-carbon units   
 Purine biosynthesis   
Raf1 MAP kinase kinase kinase −1.59 −1.35 
 Functions downstream of Ras   
 Regulation of proliferation, differentiation and apoptosis   
H2A histone family, member O Chromosome organization and biogenesis 2.23 1.44 
Sialyltransferase 1 Catalyses the transfer of sialic acid −1.36 −1.66 
 Cell surface carbohydrate determinants and antigens.   
Integral membrane protein 2B Pro-apoptotic molecule which interacts with Bcl-2 2.33 1.52 
 Mutations in this gene cause familial British dementia   
Lipin 1 Unknown −2.45 −2.05 
Immunoglobulin superfamily, member 3 Unknown −4.72 −2.38 
Gene
 
Function
 
Lymphoblast FC*
 
Osteosarcoma FC*
 
Aldose reductase Catalyses the reduction of a number of aldehydes 1.62 1.57 
 Associated with retinopathy, neuropathy and microangiopathy in diabetes   
Scaffold protein TUBA Functions to bring together dynamin with actin regulatory proteins −2.22 −1.58 
Methylenetetrahydrofolate dehydrogenase (NADP+ dependent) N10-Formyl tetrahydrofolate biosynthesis −1.38 −1.57 
 Needed for reactions requiring addition of 1-carbon units   
 Purine biosynthesis   
Raf1 MAP kinase kinase kinase −1.59 −1.35 
 Functions downstream of Ras   
 Regulation of proliferation, differentiation and apoptosis   
H2A histone family, member O Chromosome organization and biogenesis 2.23 1.44 
Sialyltransferase 1 Catalyses the transfer of sialic acid −1.36 −1.66 
 Cell surface carbohydrate determinants and antigens.   
Integral membrane protein 2B Pro-apoptotic molecule which interacts with Bcl-2 2.33 1.52 
 Mutations in this gene cause familial British dementia   
Lipin 1 Unknown −2.45 −2.05 
Immunoglobulin superfamily, member 3 Unknown −4.72 −2.38 
*

FC = fold change.

LHON-specific upregulation of aldose reductase

The upregulation of the aldose reductase transcript in cells bearing LHON mutations was of interest, as increased expression of this gene causes increased risk for optic neuropathy and RGC degeneration in the context of diabetic retinopathy (Kinoshita and Nishimura, 1988; Vinores et al., 1988). Quantitative RT–PCR results confirmed the upregulation of aldose reductase expression in LHON osteosarcoma cybrids as compared with control cell lines (Fig. 2). The upregulation of aldose reductase observed in cells bearing LHON mutations is significant (P = 0.02), and is on the order of magnitude observed in patients with diabetic retinopathy (Vinores et al., 1988; Kicic and Palmer, 1996; Shah et al., 1997; Shimizu et al., 2000).

Fig. 2

Confirmation of upregulation of aldose reductase. Expression by quantitative RT–PCR confirms an upregulation of aldose reductase in cells bearing LHON mutations. Aldose reductase expression was normalized to β-actin expression. Expression of aldose reductase in LHON cell lines was significantly altered (P < 0.05) compared with that in unfused controls, or control cybrid cell lines. Error bars represent 2SEM.

Fig. 2

Confirmation of upregulation of aldose reductase. Expression by quantitative RT–PCR confirms an upregulation of aldose reductase in cells bearing LHON mutations. Aldose reductase expression was normalized to β-actin expression. Expression of aldose reductase in LHON cell lines was significantly altered (P < 0.05) compared with that in unfused controls, or control cybrid cell lines. Error bars represent 2SEM.

Increased mitochondrially targeted aldose reductase in LHON cybrids

Recently, aldose reductase has been shown to have a mitochondrial localization, which is driven by activation of protein kinase C (Varma et al., 2003). Although we did not observe a significant increase in whole-cell aldose reductase protein in mutants, we did observe a significant increase in aldose reductase levels in mitochondrial lysates (P = 0.02), by an average of 35% in LHON cell Lines (Fig. 3A).

Fig. 3

Increased mitochondrial aldose reductase in LHON cybrids. (A). Top: western blot of aldose reductase in mitochondrial lysates from three control cybrid cell lines (lanes 1–3, H1959, HGA13 and HPC7, respectively) and four LHON cybrid cell lines (lanes 4–7, HFF3, HPE9, HMM5 and RJ206, respectively). Lane 8 contains pure human recombinant aldose reductase (AR) loaded as a positive control. Bottom: the blot above was stripped and then reprobed with a cytochrome c-specific antibody as a protein loading control. (B) Average density of the aldose reductase band from three independent blots for three control cybrid and four LHON cybrid cell lines. Error bars represent 2SEM.

Fig. 3

Increased mitochondrial aldose reductase in LHON cybrids. (A). Top: western blot of aldose reductase in mitochondrial lysates from three control cybrid cell lines (lanes 1–3, H1959, HGA13 and HPC7, respectively) and four LHON cybrid cell lines (lanes 4–7, HFF3, HPE9, HMM5 and RJ206, respectively). Lane 8 contains pure human recombinant aldose reductase (AR) loaded as a positive control. Bottom: the blot above was stripped and then reprobed with a cytochrome c-specific antibody as a protein loading control. (B) Average density of the aldose reductase band from three independent blots for three control cybrid and four LHON cybrid cell lines. Error bars represent 2SEM.

Increased sorbitol in cells bearing LHON mutations

Aldose reductase reduces glucose and galactose to sorbitol and galactitol, respectively; both of these metabolites of aldose reductase are membrane insoluble and can accumulate in diabetic patients, contributing to diabetic complications such as retinopathy and optic neuropathy (Robison et al., 1989). Our results indicate that LHON cells overexpress aldose reductase in mitochondria. We used a specific enzymatic assay based on SDH activity to measure the levels of sorbitol in control and LHON cell lines. A significant increase in sorbitol levels was found in LHON cybrids relative to controls (Fig. 4, P < 0.05 for LHON cybrids versus control cybrids, Student's t test).

Fig. 4

Increased sorbitol in LHON cells. Sorbitol levels were measured in unfused control (143B.TK), cybrid controls and LHON cybrids. Results shown are the average of five independent experiments, using one unfused control cell line, three control cybrid cell lines and four LHON cybrids (two bearing the 3460 mutation and two bearing the 11778 mutation). Error bars represent 2SEM.

Fig. 4

Increased sorbitol in LHON cells. Sorbitol levels were measured in unfused control (143B.TK), cybrid controls and LHON cybrids. Results shown are the average of five independent experiments, using one unfused control cell line, three control cybrid cell lines and four LHON cybrids (two bearing the 3460 mutation and two bearing the 11778 mutation). Error bars represent 2SEM.

Discussion

Cybridization produces a stable transcriptomal signature that persists even after mitochondrial repopulation

Because of the ability to insert mutant mtDNA into cells with a common nuclear background, microarray of cybrid osteosarcoma cells would seem to be the optimal model system for transcriptomal analysis of the consequences of pathogenic mtDNA mutations. However, by comparison of unfused controls, control cybrids and LHON mutant cybrids, we demonstrated that there are significant, large and stable consequences of the cybridization process, even after repopulation of rho-zero cells by both mutant and normal mitochondrial genomes. By comparison of unfused controls and control cybrids and rho-zero cells, we were able to infer that the majority (62%) of the cybridization-dependent transcriptomal alterations resulted from the rho-zero process itself.

The cybridization-dependent signal is mitochondrial, and preferentially upregulates OXPHOS genes

As demonstrated by the Onto-Express analysis, the most significant cybrid-dependent perturbation of the transcriptome is in nuclear genes expressed in the mitochondrial compartment, as such genes in aggregate are altered by a deviation from random expectation that is four orders of magnitude more significant than the next most affected compartment (Table 1). The mitochondrial genes affected are preferentially those involved in electron transport and ATP synthesis (i.e. OXPHOS), and the significant upregulations outweigh the downregulations by a ratio of 9 : 1 (Table 2). The simplest explanation of this observation is that the nucleus senses a mitochondrial deficiency that occurs during the long, slow depletion of mtDNA using the nuclear and mitochondrial mutagen EtBr, and attempts to adapt to it by upregulating mitochondrially targeted genes, many of which are involved in OXPHOS. This upregulation appears to persist even after the fusion step, i.e. mitochondrial repopulation. The simplest hypothesis to explain the persistence of the upregulated mitochondrial genes after repopulation is that during the mitochondrial depletion, an activation of a mitochondrial-specific transcription factor occurs, or mutation and selection for increased transcription of multiple mitochondrially targeted genes.

Potential effects of transformation on our observations

Both the osteosarcoma cells and the lymphoblasts are transformed, the former during their original isolation from the patient and the latter by Epstein–Barr virus (EBV) transfection. However, we believe that the large and mitochondrial-specific effect of cybridization on the transcriptome is explained much more simply by cybridization than transformation, for the following reasons. First, all cells we have compared have undergone transformation, so transformation itself is not different in our mutant versus control comparison. Secondly, nuclear-encoded mitochondrial genes are distributed on many nuclear chromosomes, so aneuploidy is an unlikely explanation. Thirdly, the largest number of stable alterations (comparison A), observed under conditions where cells are experiencing a long, slow depletion of the mitochondrial genome, whose 37 gene products interact intimately with nuclear-encoded ones, is mediated by the potent nuclear mutagen EtBr. Lastly, we have focused on those mutations that are shared among multiple mutant versus control comparisons between cell types, so that the results should depend on mtDNA mutation status, which is different between the groups compared, and not transformation status, which is common to all groups compared.

Other microarray studies of rho-zero and mitochondrial cybrids

Mitochondrial influence on nuclear gene expression has been examined previously in cell lines (thyroid and breast cancer) depleted of their endogenous mtDNA through treatment with EtBr, and then assaying for differential expression of nuclear genes by cDNA microarray and differential display RT–PCR (Thomas et al., 1999; Delsite et al., 2002). A direct comparison of the previously published cDNA microarray experiment on rho-zero cells done by Delsite et al. (2002) and ours is difficult, given that the former was carried out on a platform that is no longer commercially available. Also, the methods of analysis were different in that Delsite et al. only reported genes with a fold change >3-fold, whereas we identified differential expressed genes based on t tests of mutant versus control values. Some similarity was observed, in that four genes were significantly altered in the same direction in both their and our results: general transcription factor IIH polypeptide 3, tubulin-specific chaperone D, CDC28 protein kinase and protective protein for β-galactosidase. However, the previous study did not use the recently developed Onto-Express program to identify the intracellular compartmentalization of the rho-zero-dependent effects, and did not report as large and significant decrease in mitochondrial transcripts, which we did detect; this could be due in part to differences in criteria and our larger study design, which separated rho-zero- and cybridization-dependent transcripts. This is the first report to our knowledge describing the permanent effects of the cybridization process by examining global gene expression.

Cell specificity and cell autonomy of the LHON signature on the transcriptome

By filtering out the cybridization-dependent genes, we identified 96 genes whose altered expression was LHON dependent at the P < 0.05 criterion. To resolve the issue of LHON dependence and cybridization independence further, we also carried out a comparison of LHON lymphoblasts and controls (comparison E). Of the of the 96 LHON-dependent genes in the osteosarcoma line, and the 715 LHON-dependent genes that were significantly altered among control and LHON lymphoblast cell lines, only nine were shared (Table 4). This was surprising, and confirms our and other's findings that there are huge effects of cell type on the transcriptome, and the notion that there are many more cell type-dependent consequences of LHON mutations than cell-autonomous ones. The limited amount of shared transcriptional alterations could also result from a too-stringent screening. For example, if a particular transcript is decreased by the rho-zero process, and decreased then further by the LHON process, we will have removed those LHON-dependent genes because they have been classified as rho-zero dependent. A broader analysis of more cell types and mitochondrial diseases suggests that transcripts altered in rho-zero cells are frequently shared in the same direction in other cell types and mitochondrial diseases (in preparation).

LHON-dependent alterations in gene expression and potential mechanisms

After comparing the alterations in gene expression in lymphoblasts (715 genes) and osteosarcoma cell lines (96 genes), we identified nine genes that were shared: aldose reductase, H2A histone and ITM2B were upregulated; and MTHFD, scaffold protein TUBA, sialyltransferase 1, Raf1, lipin 1 and immunoglobulin superfamily, member 3 were downregulated. The mechanisms by which these may relate to LHON are described below.

Upregulation of aldose reductase in LHON

While LHON and diabetic retinopathy are clinically and pathologically very distinct, a common ground is represented by the occurrence of optic nerve degeneration and impaired axoplasmic transport (Masanori, 1998, 2000). In diabetic retinopathy, the accumulation of the membrane-insoluble metabolites of aldose reductase has been associated with cataracts and microvascular complications (Robison et al., 1989; Lorenzi and Gerhardinger, 2001; Sheetz and King, 2002). Aldose reductase expression has been shown to be induced by oxidative stress, and specifically lipid peroxides, and has been reported to metabolize the products of lipid peroxidation; and is also induced by osmotic stress (Kinoshita and Nishimura, 1988; Spycher et al., 1997; Choudhary et al., 2003; Galvez et al., 2003). We previously have observed that LHON mutations result in increased reactive oxygen species (ROS) production, that is specifically inhibited by rotenone (Wong et al., 2002). Thus, increased ROS production in neural cells bearing LHON mutations could explain both an increased mitochondrial aldose reductase expression and increased sorbitol levels that we detected in LHON cells. In this scenario, LHON mutations induce increased production of ROS which upregulates the expression of aldose reductase. The increased aldose reductase drives the subsequent increase in its enzymatic products which accumulate, potentially causing osmotic stress and toxicity to retinal ganglion cells, ultimately activating the apoptotic cascade.

LHON cybrids are both deficient in growth on galactose, which is metabolized by aldose reductase, and also die by a caspase-independent mechanism of cell death after switching from glucose to galactose medium (Ghelli et al., 2003; Zanna et al., 2003). These same alterations in galactose metabolism have not been reported in cybrids from some other mitochondrial diseases such as mitochondrial encephalomyopathy, lactic acidosis and stroke-like symptoms (MELAS), myoclonic epilepsy and ragged red fibres (MERRF) or neurogenic muscle weakness, ataxia and retinitis pigmentosa (NARP) (Hofhaus et al., 1996; Ghelli et al., 2003); thus it may be that the alteration in galactose metabolism is a LHON-dependent rather than mitochondrial disease-dependent phenotype. Our observation of an upregulation of aldose reductase may provide the underlying basis for these previous findings, in that an increase in aldose reductase is expected to cause an increase in galactitol, a toxic metabolic dead-end, and so cells overexpressing it should accumulate more of this toxic product especially in galactose media, and die.

Inhibition of Ras-dependent transcripts in LHON cells: Raf1, sialyltransferase and TUBA

Multiple transcripts whose gene products function downstream of Ras are downregulated in a LHON-dependent way, which includes Raf1, sialyltransferase and TUBA. Ras is a member of the small GTPase superfamily, and is involved in the regulation of many cellular processes including proliferation, differentiation and apoptosis (Malumbres and Pellicer, 1998). Raf1 is a downstream target of Ras; the activation of Raf1 normally induces cell survival, partially through the inactivation of pro-apoptotic factors such as Bad (Wang et al., 1996). Additionally, it has been demonstrated that osmotic stress inhibits Raf1 activity, induces caspase-3 activity and causes mitochondrial fragmentation (Copp et al., 2004). This suggests the following possible LHON mechanism: ROS→aldose reductase induction→osmotic stress→(Ras)→inhibition of Raf1→apoptosis. A predisposition of LHON cells for caspase-dependent and -independent death has been noted (Danielson et al., 2002; Ghelli et al., 2003; Zanna et al., 2003), as well as a decreased ability of neural cells bearing LHON mutations to differentiate (Wong et al., 2002).

Sialyltransferase 1 (Siat1) which is downregulated in LHON has been demonstrated to be regulated by Ras (Dalziel et al., 2004). Ena/VASP and N-WASP which are direct binding partners of TUBA (downregulated in LHON) are also regulated by the Ras pathway (Nur et al., 1999; Legg and Machesky, 2004).

Integral membrane protein 2B/BRI

ITM2B is a pro-apoptotic factor which interacts with mitochondria through its contacts with Bcl-2, and association of ITM2B with mitochondria correlates with decreased mitochondrial membrane potential and release of cytochrome c through an unknown mechanism (Fleischer et al., 2002). The observation that LHON mutations associate with and increase in ITM2B suggests another potential molecular basis for the increased apoptosis we and others have observed in LHON cells (Danielson et al., 2002; Ghelli et al., 2003). Mutational extensions of ITM2B (also known as BRI) appear to cause the disease familial British dementia (Akiyama et al., 2004).

Methylenetetrahydrofolate dehydrogenase

MTHFD1, whose transcript is decreased in LHON, plays a critical role in converting folate, a required dietary nutrient, into N10-formyl tetrahydrofolate, which is then used in numerous biochemical reactions involving the addition of 1-carbon units. Folate deprivation in humans can lead to central vision loss (Carelli et al., 2002a). Interestingly, folate appears to help specifically in the repair of damaged adult RGCs, which are primarily affected in LHON (Iskandar et al., 2004). Thus, it could be particularly important for those at risk for LHON to not become folate deficient.

H2A histone family, member O

Histones comprise a family of basic proteins that are responsible for organizing genomic DNA in eukaryotes. It is interesting to note that during apoptosis, cytochrome c induces the release of acetylated histone 2A specifically into the cytosol (Nur et al., 2004); the mechanism behind this release and its implication for apoptosis or LHON is unclear.

Overarching LHON hypothesis

Thus the additive effects of increased ROS detected in LHON cells (Wong et al., 2002), with osmotic stress caused by increased expression of aldose reductase, which results in decreased expression of pro-survival and proliferation/differentiation factors (Raf1 and Siat1), combined with the increased expression of pro-apoptotic genes (ITM2B) and the alterations in tetrahydrofolate biosynthesis (MTHFD1) may predispose RGCs to apoptosis, resulting in the specific cell death observed in LHON patients and in cell lines which bear LHON mutations (Danielson et al., 2002; Ghelli et al., 2003).

Inhibition of aldose reductase as a potential therapy

Several inhibitors of aldose reductase exist and many have been used clinically to treat patients with diabetic retinopathy. Thus, depending on the relative contribution of aldose reductase and production of its metabolites to the overall pathology of LHON, the use of aldose reductase inhibitors may become an attractive treatment strategy for a disease for which there is no current therapy available.

We wish to thank Dr Alice Wong for her helpful comments. This work was supported by United States Public Health Services grants EY12245, AG11967, AG16719, AG23311 (to G.A.C.), and by Telethon Fondazione Onlus Italy grant GGPO2323 (to V.C.).

References

Akiyama H, Kondo H, Arai T, Ikeda K, Kato M, Iseki E, et al. Expression of BRI, the normal precursor of the amyloid protein of familial British dementia, in human brain.
Acta Neuropathol (Berl)
 
2004
;
107
:
53
–8.
Barrientos A, Moraes CT. Titrating the effects of mitochondrial complex I impairment in the cell physiology.
J Biol Chem
 
1999
;
274
:
16188
–97.
Bergmeyer HU, Gruber W, Gutman I. d-Sorbitol assay.
Methods Enzym Anal
 
1974
;
4
:
1323
–30.
Brown MD, Trounce IA, Jun AS, Allen JC, Wallace DC. Functional analysis of lymphoblast and cybrid mitochondria containing the 3460, 11778, or 14484 Leber's hereditary optic neuropathy mitochondrial DNA mutation.
J Biol Chem
 
2000
;
275
:
39831
–6.
Carelli V, Ross-Cisneros FN, Sadun AA. Optic nerve degeneration and mitochondrial dysfunction: genetic and acquired optic neuropathies.
Neurochem Int
 
2002
;
40
:
573
–84.
Carelli V, Vergani L, Bernazzi B, Zampieron C, Bucchi L, Valentino M, et al. Respiratory function in cybrid cell lines carrying European mtDNA haplogroups: implications for Leber's hereditary optic neuropathy.
Biochim Biophys Acta
 
2002
;
1588
:
7
–14.
Carelli V, Ross-Cisneros FN, Sadun AA. Mitochondrial dysfunction as a cause of optic neuropathies.
Prog Retin Eye Res
 
2004
;
23
:
53
–89.
Choudhary S, Srivastava S, Xiao T, Andley UP, Srivastava SK, Ansari NH. Metabolism of lipid derived aldehyde, 4-hydroxynonenal in human lens epithelial cells and rat lens.
Invest Ophthalmol Vis Sci
 
2003
;
44
:
2675
–82.
Copp J, Wiley S, Ward MW, Van Der Geer P. Hypertonic shock inhibits growth factor receptor signaling, induces caspase 3 activation, and causes reversible fragmentation of the mitochondrial network.
Am J Physiol Cell Physiol
 
2005
;
288
:
403
–15.
Dalziel M, Dall'Olio F, Mungul A, Piller V, Piller F. Ras oncogene induces beta-galactoside alpha2,6-sialyltransferase (ST6Gal I) via a RalGEF-mediated signal to its housekeeping promoter.
Eur J Biochem
 
2004
;
271
:
3623
–34.
Danielson SR, Wong A, Carelli V, Martinuzzi A, Schapira AH, Cortopassi GA. Cells bearing mutations causing Leber's hereditary optic neuropathy are sensitized to Fas-induced apoptosis.
J Biol Chem
 
2002
;
277
:
5810
–5.
Delsite R, Kachhap S, Anbazhagan R, Gabrielson E, Singh KK. Nuclear genes involved in mitochondria-to-nucleus communication in breast cancer cells.
Mol Cancer
 
2002
;
1
:
6
.
Fleischer A, Ayllon V, Dumoutier L, Renauld JC, Rebollo A. Proapoptotic activity of ITM2B(s), a BH3-only protein induced upon IL-2-deprivation which interacts with Bcl-2.
Oncogene
 
2002
;
21
:
3181
–9.
Gajewski CD, Lin MT, Cudkowicz ME, Beal MF, Manfredi G. Mitochondrial DNA from platelets of sporadic ALS patients restores normal respiratory functions in rho(0) cells.
Exp Neurol
 
2003
;
179
:
229
–35.
Galvez AS, Ulloa JA, Chiong M, Criollo A, Eisner V, Barros LF, et al. Aldose reductase induced by hyperosmotic stress mediates cardiomyocyte apoptosis: differential effects of sorbitol and mannitol.
J Biol Chem
 
2003
;
278
:
38484
–94.
Ghelli A, Zanna C, Porcelli AM, Schapira AH, Martinuzzi A, Carelli V, et al. Leber's hereditary optic neuropathy (LHON) pathogenic mutations induce mitochondrial-dependent apoptotic death in transmitochondrial cells incubated with galactose medium.
J Biol Chem
 
2003
;
278
:
4145
–50.
Hayashi J, Ohta S, Kikuchi A, Takemitsu M, Goto Y, Nonaka I. Introduction of disease-related mitochondrial DNA deletions into HeLa cells lacking mitochondrial DNA results in mitochondrial dysfunction.
Proc Natl Acad Sci USA
 
1991
;
88
:
10614
–8.
Hofhaus G, Johns DR, Hurko O, Attardi G, Chomyn A. Respiration and growth defects in transmitochondrial cell lines carrying the 11778 mutation associated with Leber's hereditary optic neuropathy.
J Biol Chem
 
1996
;
271
:
13155
–61.
Howell N, Bindoff LA, McCullough DA, Kubacka I, Poulton J, Mackey D, et al. Leber hereditary optic neuropathy: identification of the same mitochondrial ND1 mutation in six pedigrees.
Am J Hum Genet
 
1991
;
49
:
939
–50.
Huoponen K, Vilkki J, Aula P, Nikoskelainen EK, Savontaus ML. A new mtDNA mutation associated with Leber hereditary optic neuroretinopathy.
Am J Hum Genet
 
1991
;
48
:
1147
–53.
Iskandar BJ, Nelson A, Resnick D, Pate Skene JH, Gao P, Johnson C, et al. Folic acid supplementation enhances repair of the adult central nervous system.
Ann Neurol
 
2004
;
56
:
221
–7.
Johns DR, Neufeld MJ, Park RD. An ND-6 mitochondrial DNA mutation associated with Leber hereditary optic neuropathy.
Biochem Biophys Res Commun
 
1992
;
187
:
1551
–7.
Khatri P, Draghici S, Ostermeier GC, Krawetz SA. Profiling gene expression using Onto-Express.
Genomics
 
2002
;
79
:
266
–70.
Kicic E, Palmer TN. Increased white cell aldose reductase mRNA levels in diabetic patients.
Diabetes Res Clin Pract
 
1996
;
33
:
31
–6.
King MP, Attardi G. Human cells lacking mtDNA: repopulation with exogenous mitochondria by complementation.
Science
 
1989
;
246
:
500
–3.
King MP, Koga Y, Davidson M, Schon EA. Defects in mitochondrial protein synthesis and respiratory chain activity segregate with the tRNA(Leu(UUR)) mutation associated with mitochondrial myopathy, encephalopathy, lactic acidosis, and strokelike episodes.
Mol Cell Biol
 
1992
;
12
:
480
–90.
Kinoshita JH, Nishimura C. The involvement of aldose reductase in diabetic complications.
Diabetes Metab Rev
 
1988
;
4
:
323
–37.
Legg JA, Machesky LM. MRL proteins: leading Ena/VASP to Ras GTPases.
Nat Cell Biol
 
2004
;
6
:
1015
–7.
Li C, Wong WH. Model-based analysis of oligonucleotide arrays: expression index computation and outlier detection.
Proc Natl Acad Sci USA
 
2001
;
98
:
31
–6.
Lorenzi M, Gerhardinger C. Early cellular and molecular changes induced by diabetes in the retina.
Diabetologia
 
2001
;
44
:
791
–804.
Malumbres M, Pellicer A. RAS pathways to cell cycle control and cell transformation.
Front Biosci
 
1998
;
3
:
d887
–912.
Mariotti C, Tiranti V, Carrara F, Dallapiccola B, DiDonato S, Zeviani M. Defective respiratory capacity and mitochondrial protein synthesis in transformant cybrids harboring the tRNA(Leu(UUR)) mutation associated with maternally inherited myopathy and cardiomyopathy.
J Clin Invest
 
1994
;
93
:
1102
–7.
Masanori I-U YH, Yamamoto M, Naka H, Kuriyama H. Structural impairments in optic nerve of diabetic rats ameliorated with aldose reductase inhibitor.
Exp Eye Res
 
1998
;
66
:
397
–401.
Masanori I-U ZL, Naka H, Kuriyama H, Yamamoto M. Polyol metabolism of retrograde axonal transport in diabetic rat large optic nerve fiber.
Invest Ophthalmol Vis Sci
 
2000
;
4
:
4055
–8.
Nur EKA, Gross SR, Pan Z, Balklava Z, Ma J, Liu LF. Nuclear translocation of cytochrome c during apoptosis.
J Biol Chem
 
2004
;
279
:
24911
–4.
Nur EKMS, Kamal JM, Qureshi MM, Maruta H. The CDC42-specific inhibitor derived from ACK-1 blocks v-Ha-Ras-induced transformation.
Oncogene
 
1999
;
18
:
7787
–93.
Robison WG Jr, Nagata M, Laver N, Hohman TC, Kinoshita JH. Diabetic-like retinopathy in rats prevented with an aldose reductase inhibitor.
Invest Ophthalmol Vis Sci
 
1989
;
30
:
2285
–92.
Sadun AA, Win PH, Ross-Cisneros FN, Walker SO, Carelli V. Leber's hereditary optic neuropathy differentially affects smaller axons in the optic nerve.
Trans Am Ophthalmol Soc
 
2000
;
98
:
223
–32; discussion 232–5.
Shah VO, Dorin RI, Sun Y, Braun M, Zager PG. Aldose reductase gene expression is increased in diabetic nephropathy.
J Clin Endocrinol Metab
 
1997
;
82
:
2294
–8.
Sheetz MJ, King GL. Molecular understanding of hyperglycemia's adverse effects for diabetic complications.
J Am Med Assoc
 
2002
;
288
:
2579
–88.
Shimizu H, Ohtani KI, Tsuchiya T, Sato N, Tanaka Y, Takahashi H, et al. Aldose reductase mRNA expression is associated with rapid development of diabetic microangiopathy in Japanese type 2 diabetic (T2DM) patients.
Diabetes Nutr Metab
 
2000
;
13
:
75
–9.
Spycher SE, Tabataba-Vakili S, O'Donnell VB, Palomba L, Azzi A. Aldose reductase induction: a novel response to oxidative stress of smooth muscle cells.
FASEB J
 
1997
;
11
:
181
–8.
Swerdlow RH, Parks JK, Cassarino DS, Maguire DJ, Maguire RS, Bennett JP Jr, et al. Cybrids in Alzheimer's disease: a cellular model of the disease?
Neurology
 
1997
;
49
:
918
–25.
Thomas AW, Majid A, Sherratt EJ, Gagg JW, Alcolado JC. Differential expression of mRNA in human thyroid cells depleted of mitochondrial DNA by ethidium bromide treatment.
Clin Sci (Lond)
 
1999
;
97
:
207
–13.
Torroni A, Petrozzi M, D'Urbano L, Sellitto D, Zeviani M, Carrara F, et al. Haplotype and phylogenetic analyses suggest that one European-specific mtDNA background plays a role in the expression of Leber hereditary optic neuropathy by increasing the penetrance of the primary mutations 11778 and 14484.
Am J Hum Genet
 
1997
;
60
:
1107
–21.
Trounce IA, Kim YL, Jun AS, Wallace DC. Assessment of mitochondrial oxidative phosphorylation in patient muscle biopsies, lymphoblasts, and transmitochondrial cell lines.
Methods Enzymol
 
1996
;
264
:
484
–509.
Varma T, Liu SQ, West M, Thongboonkerd V, Ruvolo PP, May WS, et al. Protein kinase C-dependent phosphorylation and mitochondrial translocation of aldose reductase.
FEBS Lett
 
2003
;
534
:
175
–9.
Vinores SA, Campochiaro PA, Williams EH, May EE, Green WR, Sorenson RL. Aldose reductase expression in human diabetic retina and retinal pigment epithelium.
Diabetes
 
1988
;
37
:
1658
–64.
Wallace DC, Singh G, Lott MT, Hodge JA, Schurr TG, Lezza AM, et al. Mitochondrial DNA mutation associated with Leber's hereditary optic neuropathy.
Science
 
1988
;
242
:
1427
–30.
Wang HG, Rapp UR, Reed JC. Bcl-2 targets the protein kinase Raf-1 to mitochondria.
Cell
 
1996
;
87
:
629
–38.
Williams AJ, Murrell M, Brammah S, Minchenko J, Christodoulou J. A novel system for assigning the mode of inheritance in mitochondrial disorders using cybrids and rhodamine 6G.
Hum Mol Genet
 
1999
;
8
:
1691
–7.
Wong A, Cavelier L, Collins-Schramm HE, Seldin MF, McGrogan M, Savontaus ML, et al. Differentiation-specific effects of LHON mutations introduced into neuronal NT2 cells.
Hum Mol Genet
 
2002
;
11
:
431
–8.
Wong A, Cortopassi G. mtDNA mutations confer cellular sensitivity to oxidant stress that is partially rescued by calcium depletion and cyclosporin A.
Biochem Biophys Res Commun
 
1997
;
239
:
139
–45.
Zanna C, Ghelli A, Porcelli AM, Carelli V, Martinuzzi A, Rugolo M. Apoptotic cell death of cybrid cells bearing Leber's hereditary optic neuropathy mutations is caspase independent.
Ann NY Acad Sci
 
2003
;
1010
:
213
–7.

Author notes

1Department of Molecular Biosciences, University of California Davis, Davis, CA, 2Department of Neurological Sciences, University of Bologna, Bologna, 3Scientific Institute Eugenio Medea, Conegliano Research Center, Conegliano, Italy, 4Department of Clinical Neurosciences, Royal Free Hospital and University College Medical School, London, UK and 5Departments of Medical Genetics and Biology, University of Turku, Turku, Finland