Cybrids in Alzheimer's disease: A cellular model of the disease?

Alzheimer's disease (AD), the most common neurodegenerative disease of late life, manifests clinically as a progressive impairment of cognitive abilities.¹ Mechanisms of cell death pertinent to the neurodegenerative diseases in general are also likely to apply in AD. Hypotheses of neurodegeneration in this category of disease emphasize that bioenergetic failure and increased production of reactive oxygen species(ROS) contribute to initiation of apoptosis, a genetically programmed process of neuronal demise.^2,3 The relevance of these hypotheses to AD is underscored by the presence of a specific bioenergetic defect in multiple tissues in AD patients. The electron transport chain enzyme (ETC) cytochrome c oxidase (complex IV; COX) is abnormal in AD brain and platelets.^4-8 Moreover, the AD COX defect acts as an ROS generator.⁹

Cytochrome c oxidase is a 13-subunit multimeric enzyme, and three of its subunits are encoded by mitochondrial DNA (mtDNA). Mutations in two of these COX subunit encoding mtDNA genes (COI and COII) were recently shown to be the basis of the AD COX defect.¹⁰ The relevance of this biochemical lesion to AD is illustrated by the fact that inhibition of COX in rats results in their cognitive impairment and abnormal hippocampal long-term potentiation.¹¹ Also, the recent demonstration of increased risk for AD in offspring of AD-affected women is consistent with this genetic model since mtDNA is maternally inherited.¹²

Mitochondrial DNA isolation from donors with subsequent transfer to immortalized cells depleted of endogenous mtDNA creates cytoplasmic hybrid(cybrid) cell lines that recapitulate at cellular and subcellular levels any specific functional abnormalities characteristic of the transferred mtDNA. The cybrid technique is useful both in demonstrating the functional consequences of known mtDNA mutations¹³ and for screening for mtDNA mutations in candidate diseases.^14-15 A cybrid tissue culture model of AD using human SH-SY5Y neuroblastoma cells exists.¹⁶ To confirm the AD neuroblastoma cybrid work in a different cell type and analyze further the consequences of the mtDNA-derived COX defect in AD, we created an mtDNA-depleted Ntera/D1 (NT2) human teratocarcinoma cell line (ρ⁰ cells) and transferred into these cells mtDNA from AD patients to form AD cybrids.

Methods. Creation of a human NT2 teratocarcinomaρ⁰ cell line. Clonal NT2 human teratocarcinoma cells were obtained from Stratagene (La Jolla, CA). This cell line was chosen because NT2 cells can be differentiated to form terminal neurons, express N-methyl-D-aspartic acid (NMDA) and non-NMDA receptors, and secrete amyloid.^17-21 Ntera/D1 cells were plated in tissue culture flasks in Optimem I medium (Gibco BRL, Gaithersburg, MD) supplemented with 10% nondialyzed,heat-inactivated fetal bovine serum (FBS) (Intergen, Purchase, NY), 1% penicillin-streptomycin, 25 ng/mL ethidium bromide, 100 µg/mL pyruvate, and 100 µg/mL uridine. We also added 300 µM DL-2-amino-5-phosphonovaleric acid and 20µM 6,7-dinitroquinoxaline-2,3 (1H,4H)-dione to block NMDA andα-amino-3-hydroxy-5-methyl-isoxazole-4-propionic acid (AMPA) receptors, and to protect cells from glutamate toxicity during the ρ⁰ creation process. Cells were incubated in darkness to avoid photodegradation of ethidium bromide, and were passaged when confluence was reached. The duration of exposure to the ethidium bromide-containing media was 120 days, at which point the resulting ρ⁰ cells were placed in Optimem I medium supplemented only with 10% heat-activated FBS, 1% penicillin-streptomycin, 100 µg/mL pyruvate, and 100 µg/mL uridine.

Preparation of cybrids. Participation of AD and control subjects as donors of mtDNA was Institutional Review Board approved. AD subjects were recruited from ongoing AD drug trials at the University of Virginia, had been diagnosed as having probable AD using NINCDS-ADRDA criteria,²² and did not manifest signs or symptoms of an alternative neurodegenerative disorder. None had familial AD. All were ambulatory outpatients. Control subjects consisted primarily of the spouses of the AD subjects.

Following informed consent, 6 mL of blood was collected in tubes containing acid-citrate-dextrose. Other anticoagulants were less suitable. The mean age of the AD group (N = 15) was 72.1 ± 2.0 years (SEM), and for the control group (N = 9) it was 65.8 ± 2.2 years (not significant). Males constituted 53% of the AD group and 56% of the control group. Platelets were used as the source of mtDNA in these experiments because they are easily obtained, contain mtDNA but not nuclear DNA, and express a COX defect and the mtDNA mutations associated with it.¹⁰ Platelets were isolated and fused as previously described with only minor modifications.²³ Three million NT2 ρ⁰ cells were harvested for each fusion. These cells were washed in fresh Optimem I medium containing 10% FBS instead of S-MEM, and the polyethylene glycol fusion step lasted 60 seconds. Following polyethylene glycol-mediated fusion of ρ⁰ cells and platelets, the cells were kept in the standard ρ⁰ cell growth medium for 1 week and then placed in a selection medium consisting of Optimem I medium; 10% dialyzed, heat-inactivated FBS; and 1% penicillin-streptomycin. Since untransformedρ⁰ cells require uridine/pyruvate for survival, untransformed cells were eliminated and only those ρ⁰ cells that had incorporated mtDNA from the donor platelets were retained. The selection process was complete after 2 weeks in this nonuridine/pyruvate-supplemented medium. Mock fusions in which ρ⁰ cells were exposed to polyethylene glycol in the absence of platelets selected completely and did not exhibit reversion to aerobic competency.

Mitochondrial DNA analysis of ρ⁰, NT2, and cybrid cell lines. Cells were harvested by trypsinization and washed in phosphate-buffered saline. Approximately 2 to 10 million cells were resuspended in 0.7 mL Tris-EDTA buffer (TE) (10 mM Tris-Cl; pH, 8.0, containing 1 mM EDTA), to which 25 µL of 10% sodium dodecyl sulfate(SDS) was added. After incubation for 10 minutes at 37 °C, 15 µL of 5 M NaCl were added and DNA was extracted three times with an equal volume of phenol/chloroform/isoamyl alcohol (25:25:1). Excess phenol was removed from the aqueous phase by extraction with an equal volume of chloroform and, after addition of 100 µL of 5M NaCl, DNA was precipitated in 1 mL ethanol. After storage at -70 °C for 1 hour, DNA was collected by centrifugation at 4 °C, washed with 200 µL 70% ethanol, air dried, and resuspended in 100 µL of TE.

Samples of DNA (0.5 µg) extracted from cybrids, NT2 cells, andρ⁰ cells were made up to a total volume of 90 µL with TE buffer. Ten microliters of 2M NaOH were added to each tube and the DNA was denatured at 65 °C for 30 minutes. At the end of this incubation period, tubes were placed on ice and 100 µL of 2M ammonium acetate were added to neutralize the NaOH. The entire 200 µL of each sample was slowly aspirated under vacuum through a membrane previously soaked in 10 × sodium citrate/NaCl (10 × SSC) and supported on three wetted filters in a dot blot manifold (Biodot, Biorad, Hercules, CA). Ten × SSC was diluted from 20 × SSC stock (3M NaCl; 0.3M sodium citrate; pH, 7.0). When all samples had been aspirated, the apparatus was disassembled and the nylon membrane was air dried. The DNA was crosslinked to the nylon membrane using an FB-UVXL-1050 UV Crosslinker (Fisher Scientific, Pittsburgh, PA). Membranes were blocked with I-Block (Tropix, Bedford, MA) at room temperature with gentle rocking for 30 minutes, rinsed quickly in hybridization buffer (5 × SSC, 1% SDS, 0.5% bovine serum albumin), and prehybridized in this buffer at 42 °C for 30 minutes with gentle rocking. Hybridization to a COI probe at a final concentration of 5 pmol/µL in prewarmed buffer at 42 °C proceeded with gentle rocking for 30 minutes. The sequence of this COI probe was 5′ CGTTTGGTATTGGGTTATGGC 3′. Membranes were washed in the following solutions: 10 minutes in 1 × SSC, 0.1% SDS; twice in 0.5 × SSC, 0.1% SDS at 52 °C for 3 minutes; 1 × SSC, 1% TritonX-100 for 3 minutes with gentle agitation; 1× SSC for 10 minutes; and two brief rinses in 50 mM NaHCO₃ and 1 mM MgCl₂ (pH, 9.5). Membranes were sealed in plastic bags in the presence of an alkaline phosphate detection agent (Lumiphos Plus, Gibco BRL, Gaithersburg, MD) and incubated for 30 minutes in the dark. Dot blots were then exposed to X-ray film (X-GMAT, Eastman Kodak, Rochester, NY) for varying times (1 hour to overnight) and the films were developed according to the manufacturer's instructions.

The cybrid lines (control, N = 4; AD, N = 5) were blotted 40 to 75 days following mitochondrial transfer and the ρ⁰ cells were blotted 7 months after the removal of these cells from ethidium bromide.

Assay of COX activity. Six to ten million cells per line were harvested in trypsin-EDTA and transferred into 15-mL conical tubes containing the standard cybrid maintenance medium (Optimem I, 10% heat-inactivated FBS, 1% penicillin-streptomycin, and 50 µg/mL uridine). Cells were washed by centrifugation to a pellet, resuspended in Hanks buffered salt solution(HBSS), and recentrifuged. The resulting whole-cell pellet was resuspended in calcium/magnesium-free HBSS. Cytochrome c oxidase activity was spectrophotometrically determined as the apparent first-order rate constant and referenced to protein as previously described.⁷

Assays were performed on native NT2 cells at the start of their exposure to ethidium bromide and at selected intervals during this exposure. After 100 days of ethidium bromide treatment, COX activity was undetectable, consistent with the generation of a ρ⁰ status. The absence of COX activity was essentially stable for at least 7 months following the removal of cells from ethidium bromide.

Cytochrome c oxidase activities in AD and control cybrid cell lines were determined between 15 and 50 days following platelet mtDNA transfer, and were found to increase gradually as repopulation of theρ⁰ cells with new mitochondria took place. After 30 days COX activities were stable. Only assays performed in cybrid lines more than 30 days old were included in the final statistical analysis of COX activities in AD and control subjects. Most cybrid lines were assayed two or more times between 31 and 50 days after PEG fusion, and the average of these individual assays represented the final COX activity for a given cybrid line. The means for the 15 AD cybrid lines and nine age-matched control cybrid lines were calculated, and these means were compared by two-tailed Student's t-test (SigmaPlot, Jandel Scientific, San Rafael, CA). Mean COX activities for AD and control cybrid lines are expressed as sec^-1/mg protein ± SEM.

Analysis of ROS. We plated 96-well microtiter plates with 80,000 cells per well (Falcon 3872 Primaria, Becton Dickinson Labware, Lincoln Park, NJ), 24 wells per cell line, and after 48 hours of growth we measured ROS production using 2′, 7′-dichlorodihydrofluorescein diacetate (DCF-DA; Molecular Probes, Eugene, OR).^14,23,24 DCF-DA provides a measurable fluorescence signal in the presence of oxygen radicals. To provide a standardized quantitation of cell numbers in each well, the nucleus-specific fluorescent dye Hoechst 33342(Molecular Probes, Eugene, OR) was added to each well at 1 µg/mL at the time of addition of DCF-DA.²⁵ Cells were loaded with the dyes for 2 hours, rinsed twice with HBSS, and placed in 200 µL of HBSS. Fluorescence scanning was performed using a Fluoroskan II fluorocytometer(ICN Biomedicals, Huntsville, AL) with excitation at 485 nm and emission at 538 nm for the DCF-DA, and with excitation at 355 nm and emission at 460 nm for Hoechst 33342. The ratio of the DCF-DA signal and the Hoechst 33342 signal was calculated for each well, and mean fluorescence per cell of each AD or control cell line was determined by averaging all 24 wells for a given cell line. The group means for the AD (N = 5) and control (N = 6) cybrid line fluorescence ratios were calculated and expressed as mean DCF-DA fluorescence/Hoechst 33342 fluorescence ± SEM. These means were compared using Student's t-test (two-tailed).

Analysis of free radical scavenging enzymes. Glutathione reductase (GRD) was measured by the method of Carlberg and Mannervik²⁶ by monitoring the oxidation of NADPH at 340 nm in the presence of oxidized glutathione. Glutathione peroxidase activity was determined by coupling it to the oxidation of NADPH by exogenous GRD(which was added in excess) and measuring absorbance at 340 nm as described by Carmagnol et al.²⁷ Total superoxide dismutase (SOD) activity was measured according to Nagi et al.²⁸ as a function of inhibition of nitroblue tetrazolium reduction by superoxide anion radical (produced by glucose oxidase in the presence of glucose). Two millimolars of potassium cyanide were used to inhibit Cu/Zn SOD and allow for the specific determination of the activity of Mn SOD. Cu/Zn SOD activity was calculated by subtracting Mn SOD activity from total SOD activity. In the SOD assays a commercial preparation of SOD was used as a standard (Sigma, St. Louis, MO). Assays were performed in duplicate. The mean of each free radical scavenging enzyme assayed was determined for AD cybrids (N = 8) and control cybrids (N = 7), and these means were compared using Student's t-test (two-tailed).

Results. The ethidium bromide concentration used in ourρ⁰ preparation medium was determined by trial and error and was the maximal concentration of ethidium bromide in which the NT2 cells could maintain a normal morphology. Without NMDA and AMPA receptor inhibition, NT2 cells could not maintain a normal morphology at 25 ng/mL ethidium bromide. After 2 weeks in ethidium bromide, NT2 cells died if not supplemented with pyruvate and uridine at the specified concentrations, which were also determined by trial and error to be the minimal amounts necessary to prevent death of NT2 cells. There was a progressive decline in NT2 COX activity during the ρ⁰ creation process, from 0.016 sec^-1/mg protein to an undetectable level after 100 days in ethidium bromide. After 7 months of continuous culture after removal from ethidium bromide, COX activity had risen to 0.0006 sec^-1/mg protein, which represented a return of activity to only 3.8% of the NT2 baseline COX activity(figure 1).

After introduction of human platelet mtDNA into NT2 ρ⁰ cells, the resulting control cybrid lines exhibited a marked increase in COX activity from 15 to 30 days post-mitochondrial transfer. After this period, control cybrid COX activity was fairly stable. Cytochrome c oxidase activity in AD cybrid lines also exhibited an increase in COX activity over this period, and although it did not appear as stable as control cybrid COX activity from 31 to 50 days posttransfer, the measured rise in COX activity was small and did not approach that of the control cybrids by day 50 (figure 2). Statistically significant differences between AD and control cybrid line COX activities were not detected for the individual epochs indicated in figure 2, probably because of the relatively small sample sizes available for each epoch. Dot blots of cybrid lines performed after COX activity had become stable showed equivalent repopulation of both AD and control cybrid mtDNA to the level of native NT2 cells. Observed differences in COX activity between AD and control cybrid groups following stabilization of COX activity were therefore unlikely to have resulted from a gene-dose effect occurring secondary to inadequate mitochondrial repopulation of the AD cybrid lines or overrepopulation of the control cybrid lines.

Between 31 and 50 days from the time of mtDNA transfer to ρ⁰ cells, COX activity in the AD cybrid lines was 0.0071 ± 0.0002 sec^-1/mg protein (N = 15). In the control cybrid lines it was 0.0084± 0.0006 sec^-1/mg protein (N = 9). Compared with control cybrids, AD cybrid COX activity was depressed by 16% (p < 0.02; figure 3). Alzheimer's disease cybrid ROS production was 0.199 ± 0.007 DCF-DAH/Hoechst fluorescence, and control cybrid ROS production was 0.176 ± 0.004 DCF-DA/Hoechst fluorescence. This represented a 12% increase in ROS in the AD cybrids (p < 0.04) relative to control cybrids (figure 4). Alzheimer's disease cybrids exhibited significant compensatory increases in several free radical scavenging enzymes, indicating that the measured increased ROS production had a substantial functional impact and that absolute ROS production in AD cybrids would have been even higher except for these protective changes (figure 5).

Discussion. Selective depletion of mtDNA but not nuclear DNA from cultured cells using ethidium bromide, a DNA-binding mutagen, is possible because its lipid solubility and widely distributed positive charge result in its preferential uptake into the negatively charged interior of mitochondria.^29-32 Depletion of mtDNA causes loss of ETC activities including COX, since 13 ETC subunits are encoded by mtDNA. Loss of ETC activity is lethal to aerobic cells (such as those of the NT2 lineage) unless special metabolic supplementation is provided. Uridine is usually required because dihydro-orotate dehydrogenase, which is required for the synthesis of pyrimidine intermediates, is coupled to the ETC.³³ Pyruvate supplementation is also required because it facilitates recycling of reducing equivalents in the cytosol and may also enter alternate bioenergetic pathways.^34,35 In addition to permitting cell survival during conversion to ρ⁰ status, the dependence of ρ⁰ cells on pyruvate and uridine provides a selection tool useful for removal of untransformed ρ⁰ cells.

Metabolic characteristics unique to a particular cell line may require modifications of this previously published mtDNA-depletion technique. Undifferentiated NT2 cells exhibit partial and limited glutamate receptor expression,¹⁸ which may explain why glutamate receptor blockade was found to help this cell line tolerate conversion to ρ⁰ status and the resultant induction of bioenergetic failure. After 100 days of exposure to ethidium bromide, achievement of functional ρ⁰ status was indicated by loss of COX activity and an absolute requirement for added pyruvate/uridine for survival. An additional incubation period of 20 days was arbitrarily selected since longer periods of ethidium bromide exposure appear to decrease spontaneous reversion of ρ⁰ cells to aerobic competency.²³ Seven months after removal from the ethidium bromide medium, only trace COX activity was evident, indicating genetic stability of the NT2 ρ⁰ cell line.

Individual cybrid lines differ only in their mtDNA, and differences between various cybrid lines must arise from differences in mtDNA. Nuclear genes cannot account for differences since the ρ⁰ cells were prepared from the same clonal parental line and nuclear genotypes are equivalent. Differences cannot arise as a result of faulty mtDNA transfer, which is performed identically for all cell lines, because measurements of mtDNA show AD and control cybrids to contain equivalent amounts of mtDNA. Differences cannot result from environmental factors since all cybrid lines were exposed to the same growth conditions. Nongenetic factors such as toxins that might be transferred with platelets should have degraded and are massively diluted after many, many cell doublings in culture. Latent but potentially confounding ethidium bromide-induced nuclear DNA damage acquired during generation of the ρ⁰ stock is controlled by comparing AD cybrid lines with control cybrid lines instead of native NT2 cells.

Cybrid COX activity initially was relatively low for both AD and control cybrid cell lines 2 to 3 weeks following exogenous mitochondrial reconstitution, and probably reflects a gradual equilibration of cell mitochondrial mass over time as the ρ⁰ cells were repopulated with exogenous mtDNA. The very gradual increase of COX activity in AD cybrids after prolonged culture probably reflects the heteroplasmic nature of the mtDNA lesion, since less impaired cells grow more quickly and crowd out the more severely impaired cells.²⁹ Consequently, our studies probably underestimate the magnitude of the true COX defect in AD. Differentiated cybrids that are no longer replicating may well exhibit larger losses of catalytic activity. The use of platelets as mtDNA donors may also contribute to an underestimation of the actual in vivo AD neuronal COX defect, since the mtDNA mutational burden within platelets may not reflect the magnitude of mutational burden within postmitotic neurons.

Alzheimer's disease cybrid COX activity was impaired when compared with age-matched control cybrid lines. This indicates the presence of mtDNA COX gene mutations in persons with AD and confirms a previous report that these mutations are sufficient to cause the COX defect present in AD patients.¹⁰ Equivalent recoveries of mtDNA between control and AD cybrid lines on dot blot analysis provide further indication that this bioenergetic defect is catalytic in nature.³⁶

As in human neuroblastoma SH-SY5Y AD cybrid lines, our human teratocarcinoma NT2 AD cybrids demonstrate increased production of ROS.⁹ This suggests that the observed catalytic defect, while small in magnitude, is nevertheless likely to have pathophysiologic consequences. The COX defect in AD constitutes a genetically determined free radical generator. Reactive oxygen species are elevated in our AD cybrids despite the fact that ROS scavenging enzymes are upregulated, presumably as a compensatory response. However, the increase in free radical scavenging enzyme activities is not sufficient to handle the elevated ROS burden generated by defective AD COX. Increased ROS production is likely to contribute to neuronal death in AD and represents an important gain-of-function phenomenon arising secondary to the mtDNA-derived COX defect of AD. Cytochrome c, a component of the ETC embedded in the inner mitochondrial membrane, has likewise been shown to be part of the apoptotic signaling pathway and we hypothesize that defective mitochondria may further contribute to apoptosis through release of cytochrome c from damaged mitochondrial membranes.³⁷

Cybrid cell lines should be useful for studying many aspects of the pathophysiology of AD and other neurodegenerative diseases. Static and dynamic calcium homeostasis is altered in Parkinson's disease cybrids, indicating that a primary mtDNA abnormality can perturb these functions.³⁸ Studies of calcium homeostasis are indicated in AD cybrids. Because differentiated NT2 cells express glutamate receptors, we can now explore how a disease-specific bioenergetic defect affects excitotoxicity thresholds.² Ntera/D1 cells also produce amyloid, and will serve as a system in which to study if and how a specific COX lesion alters cell amyloid metabolism.³⁹ Ntera/D1 AD cybrids provide an in vitro model for evaluating the ability of drugs to ameliorate specific disease processes. For example, our current findings suggest agents with ROS scavenging qualities or the ability to bolster free radical defenses warrant efficacy studies.

The observation that a primary biochemical defect (COX dysfunction) gives rise to a secondary biochemical defect (increased ROS generation) raises the issue of whether a quantifiable correlation exists between these parameters. Ideally, the cybrid system should lend itself to such an analysis, but in reality it may not provide the considerable sensitivity required for this. In our AD cybrids the magnitude of the catalytic COX defect for a given line did not significantly correlate with ROS generation. Similarly, measured ROS generation for a given AD cybrid line did not correlate significantly with that cybrid line's free radical scavenging enzyme activities. We also did not observe a correlation between the degree of COX dysfunction and the age or duration of symptoms of the individual AD platelet donors. Unfortunately it is not clear whether failure to demonstrate such correlations reflects their absence, inherent limitations of our cybrid system, or limitations of the assay techniques used.

Generation of NT2 ρ⁰ cells with subsequent reconstitution of mtDNA from human donors results in unique cybrid cell lines that express the characteristics of the transferred mtDNA. After a brief post-transfer period, AD cybrids demonstrate a stable defect of COX catalytic activity, indicating mtDNA mutations are sufficient to cause a bioenergetic defect that is present in multiple tissues of persons with this disease. Reactive oxygen species levels are elevated in AD cybrids despite compensatory upregulation of free radical scavenging enzymes. Increased ROS production therefore represents a gain-of-function consequence that arises ultimately from mutations of mtDNA. The COX defect and elevated ROS observed in our AD cybrids are likely relevant to neurodegenerative pathology, and the NT2 cybrid system should prove useful for further studies of AD pathophysiology at the cell level and for AD drug screening.

Acknowledgments

The authors wish to thank Paula Damgaard, Cindy Barnhill, Anisa Murray, George Hanna, and Robert Brashear of the University of Virginia Department of Neurology for providing blood samples from AD and control patients. Dr. Fred Wooten reviewed the manuscript.