The Biological Foundation of the Genetic Association of TOMM40 with Late-onset Alzheimer’s disease

1. INTRODUCTION

Mitochondria are semi-autonomous organelles responsible for energy production, ion homeostasis, lipid and amino acid metabolism, self-renewal and apoptosis, and their health is critical for cell survival. Mitochondrial dysfunction is a fundamental characteristic of neurodegenerative diseases, including Alzheimer’s disease[1–4]. In AD, mitochondrial dysfunction precedes detectable amyloid pathology in humans[5–7] and animal models[8, 9]. The origin and underlying cause of mitochondrial dysfunction in AD has not been identified, but it has been linked to altered expression of bioenergetics genes[10, 11], exposure to environmental toxins[12–14], mutations in mitochondrial DNA[15, 16], oxidative damage[17], and accumulation of ApoE ε4 fragments[18–20] and/or beta-amyloid peptides in the mitochondrial matrix[21–26]. Mitochondria possess approximately 1500 proteins, and nuclear genes encode greater than 95% of these, which are synthesized on cytosolic ribosomes, and are imported through the TOM (Translocase of the Outer Membrane) apparatus; Tom40 is the central pore of this apparatus, the gateway for protein entrance into the mitochondrion (reviewed in[27]). Beta-amyloid peptides and mis-directed amyloid precursor protein interfere with mitochondrial protein import and disrupts mitochondrial function[28–31]. These results suggest alterations in the function of the TOM apparatus and mitochondrial protein import contribute to Alzheimer’s disease pathology.

Late-onset, sporadic Alzheimer’s disease (LOAD; age of onset ≥ 65 years) is the most prevalent form of Alzheimer’s disease, and the most significant risk factors are age and genetics, including specific cis-haplotypes of TOMM40 and APOE. Among the APOE isoforms, APOE ε4 has the greatest risk[32], but APOE ε3 also contributes significantly[33]. A variable length deoxythymidine homopolymer polymorphism (rs10524523, hereinafter ‘523’) in intron 6 of the TOMM40 gene (chr19:45,403,049 – 45,403,083, human genome reference assembly GRCh37/hg19), makes significant contributions to LOAD genetic risk[34] and age of onset[35–38], and explains the genetic contributions of APOE ε3-containing chromosomes to LOAD risk. In Caucasians, ‘523’ exhibits three predominant length polymorphisms, “short” (T<18 residues), “long” (18<T<30) and “very long” (T>31). “Long” poly-T lengths segregate cis to APOE ε4 in almost all instances, and “short” and “very long” poly-T lengths associate with APOE ε3. In APOE ε3/ε4 carriers, the “Short” poly-T length is protective, since onset age was ca. 7 years older in individuals who carried one copy of the “S” ‘523’ allele than in those who carried one copy of the “VL” allele[36, 39]. The 523’ VL poly-T was associated with impaired cognition[40–42], reduced grey matter volume in brain regions susceptible to AD pathology in cognitively normal, late middle-aged individuals[40], and with impaired spatial navigation and reduced grey matter volumes in individuals with amnestic mild cognitive impairment[43]. TOMM40, but not APOE, contributed to worsening test-retest performance at ages <60 years, while APOE alone contributed to aging at >60 years[44]. The TOMM40 effect was attributable to the VL poly-T length[44]. The VL isoform was also associated with increased rate of vocabulary decline in a study of healthy cognitive aging[45]. These genetics analysis show TOMM40 and APOE make separate and distinguishable contributions to cognitive aging, and TOMM40, specifically the VL poly-T length, is involved at some early step in cognitive decline, and complement the conclusion of the cell biological analysis, that altered mitochondrial protein import contributes to AD pathophysiology. In contradistinction to these findings, we have also reported that among APOE ε3/ε3 homozygotes, individuals who are also homozygous for the S poly-T develop LOAD at an earlier age than S/VL heterozygotes[37].

We[46] and others[45, 47, 48] investigated the effects of the ‘523’ poly-T length polymorphisms on TOMM40 and APOE transcription. Using brain samples from APOE ε3/3 homozygotes to distinguish TOMM40 effects from those due to APOE, we found both APOE and TOMM40 mRNA levels were dose-dependent on the numbers of VL strands in brain samples from cognitively normal individuals[46]. The results of experiments with a luciferase reporter system confirmed that the VL poly-T resulted in greater expression than the S poly-T[46]. Bekris et al. also demonstrated the TOMM40 ‘523’ poly-T has a significant effect on expression of TOMM40 mRNA[47], and Payton et al. reported that the ‘523’ S poly-T length acted as a transcriptional repressor of luciferase model constructs[45].

These data suggest altered expression of TOMM40 plays a role in the cognitive and neuroimaging effects associated with the ‘523’ poly-T. Analysis of genetic constructs provides insight to how TOM40 variation might contribute to AD pathophysiology. Although homozygous TOM40 deletions are lethal in yeast[49], fungi[50] and mammals[51], hemizygous knock-down mice (created on a C57Bl/6 background) exhibited normal development and survived until adulthood, notwithstanding they exhibited 30% greater mortality than wild-type mice[51]. The hemizygotes exhibited subtle cardiac, respiratory, thermoregulatory and neurological phenotypes that worsened with age, and older mice did not survive narcotic anesthesia[51]. Respiration in brain and heart mitochondria was reduced despite similar composition, amounts and enzyme activities of the respiratory complexes in hemizygote and wild-type mice[51], and the mitochondrial profiles in electron microscopy sections of heart tissue were distorted and possessed indistinct cristae, also consistent with reduced oxidative phosphorylation[52]. Aside from slightly worse performances in social discrimination tests, the hetrozygote mice did not exhibit behaviors or learning or memory deficits associated with Alzheimer’s disease, possibly because they did not survive long enough for the phenotype to develop. The heterozygotes did develop motor defects and older heterozygotes had fewer tyrosine-hydroxylase positive neurons in the substantia nigra than comparably aged, control mice. These results support the hypothesis that altering TOMM40 mRNA abundance alters mitochondrial function, which has pathophysiological consequences, including neurological defects.

We performed the experiments described in this report to model the VL effect on TOM40 expression, and to determine how increased levels of TOM40 protein affect mitochondrial function. Previously, Hedskog et al. found no differences between the S/L and VL/L ‘523’ genotypes regarding the levels of TOM40, ApoE or PSEN2 proteins, mitochondrial membrane potential, or the mitochondrial area or profile as revealed by electron microscopy. However, their analysis was influenced by the fact TOMM40 mRNA expression did not differ between the S/L and VL/L genotypes. In addition to the potentially confounding effect of the presence of an APOE ε4 allele in each sample, genotypic differences in the regions outside the TOMM40-APOE LD block also may have affected their results. To avert this potential problem, we used a HeLa over-expression system, providing a common genomic background. Although the HeLa APOE/TOMM40 haplotype is APOE ε3/4 S/VL, HeLa cells do not express measurable quantities of ApoE protein[53], and the use of a plasmid DNA expression system obviated potential ‘523’ poly-T- and APOE ε4-mediated gene regulatory effects[54].

2. EXPERIMENTAL METHODS

2.1 Materials

Phenol red-free DMEM (11054-020, 5.55 mM glucose; 31053-028, 25 mM glucose), phenol red-free MEM (51200-038), Hank’s balanced saline solution (14025092), Earl’s balanced saline solution (24010043), pyruvate (11360-070), glutamax (35050-061), non-essential amino acids (11140-050), penicillin/streptomycin (Pen/Strep; 15140-163) and Geneticin (10131-035) were from Gibco Life Technologies. Characterized fetal bovine serum (SH30071.03) was from HyClone, fatty acid-free BSA was from Armor and trypsin was from Sigma-Aldrich. T75 flasks were from Corning, 145 mm dishes were from Greiner (639160), and 100×20 mm, 35×10mm and 60×15mm dishes, and all other routine plastic ware used for cell culture were from Falcon. Glass cover slips (#1.5, 12 mm) were from Warner Instruments, low protein binding microfuge tubes were from Eppendorf, and pipet tips were from Neptune and Molecular Bioproducts. Black-wall, Krysal^™ bottom 96-well plates were from Phenix (MP5005) and opaque white 96-well plates were from Greiner (655075). Criterion TGX 4–15% Stain Free SDS-PAGE gels (5678085), and 12% Bis-Tris gels (345-0119) were from BioRad. NativePage Novex^™ 4–16% Blue Native gels (BN1002BOX) and 4X sample buffer (BN2003) were from ThermoScientific. Immobilon PVDF membranes were from Millipore (IPVH00010) and the ECL Advance Western Blotting detection kits were from General Electric (RPN2232). MitoTracker Orange (M7510), MitoTracker Green (M7514), tetramethylrhodamine methyl ester (TMRM; T668), carboxymethyldichlorofluorofluorescin diacetate (CM-H2DCFH-DA; C6827), and ProLong Gold anti-fade reagent with DAPI (P36941) were from ThermoScientific. MitoXpress oxygen probe and high sensitivity mineral oil were from Cayman (600800). CellTiterGlo ATP detection kits were from Promega (G7571). Cyto-ID Autophagy detection kits were from Enzo (ENZ-51031). The XTT Cell Proliferation kit was from ATCC (30-1011K). PhosBlock phosphatase inhibitor (04906837001) and mini complete protease inhibitor without EDTA (05892791) were from Roche. The bicinchoninic acid protein assay kit (23229) was from ThermoFisher Scientific; all other chemicals were reagent grade or better and were from Sigma-Aldrich or Fisher.

2.2 Brain tissue

Brain samples of cognitively normal APOE ε3/ε3 individuals and APOE ε3/ε3 and APOE ε3/ε4 subjects with Alzheimer’s disease were obtained from the Joseph and Kathleen Bryan Brain Bank at Duke University, the Brain and Tissue Bank for Developmental Disorders at the University of Maryland, and the Layton Aging and Alzheimer’s Disease Center at Oregon Health and Science University, and have been described previously[55]. Tissue fragments were pulverized under liquid nitrogen, and the powder was suspended 1:10 in homogenization buffer (0.1 M NH4HCO3 pH 8.0/1 mM EDTA/1 mM EGTA/complete protease inhibitor without EDTA), and sonicated for 3 seconds at setting 3.5 in a Diagenode sonicator. Aliquots of the homogenates were snap-frozen in liquid nitrogen and stored at −80°C until use.

2.3 Cells

TOM40-expressing HeLa cells (TOM40-1, TOM40-2, and TOM40-3 cells) and matching control cells tranfected with vectors that lacked the TOM40 coding sequence (HeLaC-1, HeLaC-2, HeLaC-3 cells) were provided courtesy of Professor Daniel L. Rock, University of Illinois, Champaign-Urbana. The full-length ORF of human Tom40 (361aa, 1086 bp) was amplified in a standard RT-PCR using cDNA from HeLa cells as a template, and the nucleotide sequence of the product was confirmed by DNA sequencing. To express stably Flag-tagged Tom40 in HeLa cells, the recombinant vector pLNCX2-Tom40 was constructed and verified by DNA sequencing. Tom40-expressing cells were selected by neomycin resistance. HeLa cells transfected with the pLNCX2 vector without the TOM40 insert served as the HeLaC controls. Cells were grown at 37°C in a humidified atmosphere of 5% CO₂/95% air, in DMEM containing either 25 mM or 5 mM glucose and supplemented with 10% fetal calf serum/1 mM pyruvate/2 mM Glutamax/100 U/mL penicillin/100 ug/mL streptomycin/200 ug/mL Geneticin. We obtained nontransfected HeLa cells via the Duke University Cell repository from ATCC (ATCC^® CRM-CCL2^™); we grew these cells in MEM (5.5 mM glucose) supplemented with FBS, non-essential amino acids/1 mM pyruvate/2 mM Glutamax/Penn-Strep, as per guidelines provided by ATCC. For all the cell lines, we replaced the medium every 48 hours and passaged the cells when cultures reached 70 – 80% confluence.

2.4 Cell fractionation

Cells were grown in 145 mm tissue culture dishes. After the cultures reached 70 – 80% confluence, the cultures were washed twice with Hank’s buffered saline solution, and the cells were then harvested by trypsinization. After quenching the trypsin with serum, the cells were washed twice by centrifugation at 500xg and resuspended in Hank’s balanced buffer solution. We resuspended the final cell pellets at a ratio of 2 × 10⁷ cells per mL in homogenization medium (HM: 220 mM mannitol/70 mM sucrose/2 mM HEPES (pH 7.4 at 4°C), supplemented with protease inhibitors (Roche)), and homogenized the cells using 20 strokes of a Potter Teflon/glass homogenizer, spinning at 2400 rpm, over a period of 6 minutes on ice. The homogenates were centrifuged for 5 minutes at 1,500xg to pellet the nuclei, and the post-nuclear supernatant (PNS) fractions were centrifuged at 10,000 × g for 10 minutes to collect the heavy mitochondrial fraction (HMF) and the post-mitochondrial supernatant fraction (S1). We centrifuged the S1 fractions at 16,000xg for 30 minutes, to produce a mixed membrane pellet (MMP) and the final supernatant (S2) fraction. All centrifugations were conducted at 4°C. All fractions were flash frozen in liquid nitrogen and stored at −80°C until use.

2.5 SDS PAGE and Western blotting analysis

To determine protein abundance, we grew cells in T75 flasks or 100 mm dishes until the cultures attained 70 – 80% confluence. We dislodged the cells from the plates mechanically using a cell scraper, and collected the cells in EBSS. After centrifugation for 4 minutes at 500xg, we discarded the supernatant fractions and froze the cell pellets, and stored them above liquid nitrogen until use. We resuspended cells and pelleted cellular fractions in Tris-buffered saline (150 mM NaCl/50 mM Tris (pH7.5)/10% glycerol/2 mM EDTA/50 mM NaF, supplemented with Roche phosphatase and protease inhibitors), and we mixed the S1 and S2 fractions described in Figure 2 1:1 with the same buffer. After determining the protein concentrations of the samples (BCA protein assay, ThermoFisher), we diluted the samples as indicated below using SDS-PAGE reducing buffer (63 mM Tris, pH 7.5/10% glycerol/0.05% bromophenol blue/2% SDS/0.9% β-mercaptoethanol), boiled the mixtures for 10 minutes and applied appropriate volumes of each sample to Criterion TGX 4–15% Stain Free gels. For each protein of interest, we conducted initial experiments with three serial dilutions of the cellular extract to determine the linear range of response; in subsequent experiments for protein quantification, we used protein concentrations that were within the linear response range. After electrophoresis we estimated the total protein amount in each lane using BioRad’s Stain-Free Technology[56] on a BioRad Gel Doc EZ Imager. We then transferred the proteins to a PVDF membrane. The membranes were blocked with 5% (w/v) milk in transfer buffer containing 10 mM Tris (pH 7.5)/500 mM NaCl/0.01% Tween 20 for one hour at room temperature. The membranes were probed with primary antibody solution (in transfer buffer supplemented with 5% milk/0.01% Tween 20) for 1 hour at RT or overnight at 4°C, followed by incubation with an HRP-conjugated secondary antibody in transfer buffer/5% milk/0.01% Tween 20 for 1 hour at RT. For determination of GPR75/HSPA9, the PVDF membranes were blocked with 3% (w/v) bovine serum albumin for 1 hour at RT and then incubated with mouse anti-HSPA9 monoclonal antibody at a dilution of 1:1000 for 90 minutes at RT, followed by mouse IgG antibody (1:2000) for 1 hour at RT. Protein signals were developed with ECL detection kits. Membranes were scanned using a Protein Simple FluorChem Q fluorometer and densities of the protein bands were measured with AlphaView FluorChem Q software version 3.2.2.0. To estimate protein abundance, the densities of the test protein bands were normalized to total protein in the lane, and the values in the TOM40 and HeLa C cells were then compared to the corresponding normalized values for non-transfected HeLa cells, as described by Taylor and Posch[57].

Figure 2. Relative TOM40 levels in HeLa cells transfected with the intact expression vector (TOM40 cells) vs in cells transfected with the vector lacking the expression sequence (HeLaC cells).

TOM40 was determined by Western blotting as described in Materials and Methods. Cells were grown to 70 – 80% confluence in T-75 flasks and dislodged from the plate mechanically using a cell scraper. The cells were suspended in Hepes-buffered Earl’s balanced saline solution and collected by centrifugation at 500xg for 4 minutes. After discarding the supernatant fluid, the cell pellets were frozen and stored above liquid nitrogen until use. The cells were extracted and SDS-PAGE and Western blotting were performed as described in Materials and Methods. Criterion TGX Stain-free gels were loaded with 6 μL/lane of serial 1:1 dilutions of cell extract. The data represent the averages (n=3) of 12, 6 and 2 μg protein loadings per well, and were corrected for total lane protein load as described in Materials and Methods. The standard used for estimating abundance was a total cell extract of non-transfected HeLa cells. The standard and all experimental cell lines were plated and grown with the same lot of media and serum and all were harvested at 70 – 80% confluence. The cell pellets were extracted on the same day using the same buffers, and were analyzed on the same gel. We obtained similar results with at least three different passage numbers from the same thaw, and from four thaws of cells. A. Western blots. Top: Tom40 bands; Lower: protein bands detected in each lane by Stain-free technology. B. Quantification of Tom40 in transfected cell lines compared with the non-transfected standard. For each cell line, the results for each dilution were calculated separately and averaged. Data are means±SEM.

2.6 Immunocytochemistry

Cells (5×10⁴/well) were plated onto 12 mm round glass cover slips in the wells of a 24-well culture dish. After the cultures had attained approximately 50% confluence, MitoTracker Orange (585 nM) was added and the cultures were incubated for 1 hour at 37°C. We then removed the medium and washed the cultures 1X with pre-warmed complete medium and 3X with pre-warmed PBS. The cells were fixed with paraformaldehyde and lysed as described in[58]. After lysis, the cells were washed 1X in PBS containing 0.05% Tween-20 (PBS-T), incubated in blocking buffer (5% BSA in 500 mM NaCl/50 mM Tris, pH 7.4) for 10 minutes, and then exposed to purified anti-TOM40 antibody (1:300 in PBS-T). After 1 hour at room temperature, the cells were exposed to secondary antibody labeled with a fluorescent dye, (Alexa Fluor-488-labeled anti-rabbit IgG (H+L)), at a ratio of 1:200, for 1 hour at RT. After washing, 5–10 μL of mounting media containing Pro-Long^® Gold Antifade reagent with DAPI was added to a standard glass slide and the coverslips were mounted onto the slide. After allowing 1–2 hours for the mounting media to solidify, we examined the slides using a Leica (Buffalo Grove, Ill) SP5 confocal microscope. The nucleus was visualized at 405 nm (blue), the mitochondria at 561 nm (orange), and TOM40 at 488 nm (green). Images were captured using Leica Application Suite Advanced Fluorescence software. Pixel counts from each channel were quantified using ImageJ software (National Institute of Health). To quantify co-localization of Tom40 and mitochondria, Pearson’s correlation coefficients were calculated using Imaris (Bitplane Scientific Software, South Windsor, CT) software, according to the protocol developed by Zinchuk et al.[59]

2.7 Antibodies

Tom40 rabbit polyclonal antibodies were from Primm Biotech, and were used at a dilution of 1:5000 for detection on Western blots. The antibodies were purified by affinity chromatography and used at a dilution of 1:300 for immunocytochemistry. Tom20 (sc-11415, 1:1000) and Tom22 (J-31, sc-101286, 1:1000) rabbit polyclonal antibodies were from Santa Cruz. Tom20 and Tom22 mouse monoclonal antibodies were from Abcam (ab56783, 1:1000, and ab57523, 1:1000, respectively). GPR75/HSPA9 mouse monoclonal antibodies were from Abcam (ab2799, 1:1000), and VDAC rabbit polyclonal antibodies were from Millipore (AB10527, 1:2000). β-actin mouse monoclonal antibodies were from Cell Signaling (D68A8, 1:5000). Goat polyclonal anti-rabbit IgG antibodies were from Cell Signaling and were used at dilutions of 1:4000 for Tom40 and VDAC and at 1:5000 for Tom20. Horse polyclonal anti-mouse IgG antibodies were from Cell Signaling and were used at a dilution of 1:2000 for Tom22 and HSPA9. Mouse monoclonal anti-HSPA9 antibodies were from abcam, and were used at 1:1000. Complex I (ab109721) and complex IV (ab109909) activity kits were from Abcam.

2.8 Enzyme activity assays

We grew cell cultures to 70 – 80% confluence in 10 cm dishes, rinsed the dishes with D-PBS, and scraped the cells into a small volume of 10 mM Tris buffer, pH 7.4. We extracted the cells and conducted the assays for α-ketoglutaric acid dehydrogenase (ab185440), complex I (ab109721) and complex IV (ab109909) using kits from Abcam, according to the manufacturer’s instructions.

2.9 Oxygen consumption

We measured oxygen consumption via phosphorescence quenching using the oxygen-sensitive porphyrin-based probe MitoXpress (Cayman 600800)[60]. We inoculated each well of a 96-well Phenix black-wall, glass-bottom plate with 12×10³ cells in 200 μL growth medium. Twenty-four hours later the medium was replaced with 200 μL/well Hepes-buffered saline solution (EBSS: 81.28 mM NaCl/25 mM HEPES (pH 7.4)/39.4 mM NaHCO₃/5.4 mM KCl/1.8 mM CaCl₂/0.91 mM NaH₂PO₄/0.81 mM MgSO₄/1% BSA) containing 5 mM sodium pyruvate, 5 mM sodium malate and 0.15 μM oxygen sensing probe. Any test substances used were added at the same time as the probe. We covered the samples with 100 μL/well mineral oil, and immediately began kinetic measurements each minute for 2 hours at 37°C, using either a SpectraMax Gemini (Molecular Devices) plate reader (380_ex/650_em nm/cutoff 630 nm) or a FluoStar Optima (BMG) in the time resolved mode (355_ex/650_em; 30μsec delay, 100μsec read time). One well of each plate was used to determine fluorescence in deoxygenated medium, by adding 65 μM ascorbate and 0.2 U/well ascorbate oxidase[61]. We repeated all experiments with at least three independent cell platings, with multiple replicates (n≥6) of each concentration of each test compound. At the end of two hours, the medium was aspirated from the wells, the wells were rinsed gently 3 times with 200 μL Dulbecco’s PBS (complete with magnesium and calcium; D-PBS) per wash, and ATP was determined as described below. Aliquots of the lysates were diluted as recommended by Promega for use in protein determinations.

Oxygen was calculated according to the equation[62]:

$${[{\mathrm{O}}_2]}_{\mathrm{t}}=\{[{[{\mathrm{O}}_2]}_{\mathrm{a}}\times {\mathrm{I}}_{\mathrm{a}}\times ({\mathrm{I}}_{\mathrm{o}}-{\mathrm{I}}_{\mathrm{t}})]/[{\mathrm{I}}_{\mathrm{t}}\times ({\mathrm{I}}_{\mathrm{o}}-{\mathrm{I}}_{\mathrm{a}})]\}$$

where [O₂]_t = oxygen concentration (μM) at time t, (O₂)_a = oxygen concentration in air-saturated buffer (257.8 mM at 37C°C[63]), I_t, I_o and I_a are fluorescent intensities at time t, in deoxygenated buffer, and in air-saturated buffer, respectively. The oxygen consumption data were corrected for protein, and the slopes of the linear portions of the curves were determined by linear regression using EXCEL.

2.10 ATP determinations

To measure basal ATP levels, we plated 12×10³ cells/well in 96-well plates, and after 24 hours replaced the growth medium with EBSS supplemented with 5 mM pyruvate/5 mM malate, plus test substances as indicated in the text and figure legends. After incubating the cultures for 2 hours, we washed the cells once with D-PBS (100 uL/well), and then added 100uL/well D-PBS plus 100 uL/well CellTiter-Glo® reagent, prepared according to the manufacturer’s instructions. After mixing on an orbital shaker in the dark for two minutes, we transferred 100uL of the cell lysates to white 96-well plates (Greiner bio-one, 655075) and read the luminescence signal on a Molecular Devices SpectraMax or on a BMG FluoStar Optima plate reader. Following these measurements, portions of the cell lysates were diluted as recommended by Promega for protein determinations.

2.11 ROS measurements

Cells were plated and grown as described above for measuring cellular respiration. After 24 hours, the growth medium was replaced with pyruvate- and malate-supplemented EBSS plus the test compounds as indicated in the text and figure legends. After 2 hours at 37°C, 10 μM 5,6-carboxy-2′,7′-dichlorodihydrofluorescin diacetate was added to each well. After 30 minutes, the medium was aspirated, each well was washed 3X with 200 μL/wash D-PBS, and end-point fluorescence measurements were made using a Molecular Devices Gemini plate reader (485_ex/538_em/530 nm cutoff) or a BMG FluoStar Optima (485_ex/530-10_em filter).

2.12 TMRM and MitoTracker Green fluorescence measurements

Cells were plated into 96-well plates and grown as described above for ATP measurements. Twenty-four hours after plating, we replaced the growth medium with EBSS supplemented with 5 mM pyruvate/5 mM malate, plus other additives as indicated in the text and table legends, and allowed the cultures to incubate for 60 minutes at 37°C. We then added TMRM (110 nM final concentration) or MitoTracker Green (150 nM final concentration). After incubating the cells for an additional 60 minutes, we washed the cultures 3X with EBSS (200μL/wash). After the final wash, we added 100 μL EBSS and measured the fluorescence using 485_ex/530-10_em filters (MitoTracker Green), or 544_ex/590_em filters (TMRM), using a BMG FluoStar Optima plate reader. We then aspirated the medium, rinsed the cells 3X with PBS without BSA, and after the final wash added 100uL/well 0.5N NaOH for 30 minutes at 37°C to lyse the cells. Portions of the lysates were used to measure protein, and fluorescence values were adjusted to total protein/well.

2.13 Autolysosome detection

To determine how TOM40 over-expression affects the extent of autophagy, we grew cells in 96-well plates as described above. Twenty-four hours later we replaced the growth medium with fresh medium with or without 10 μM FCCP. After 12 hour incubation, we replaced the growth medium with EBSS containing 5 mM pyruvate and 5 mM malate, added 60 μM chloroquine to positive control wells that had not been exposed to FCCP. We incubated the plates for 4 hours, and determined autolysosome staining using the Cyto-ID autophagy detection kit, as described by the manufacturer.

2.14 XTT test for cell viability

We plated cells at a density of 20×10³ cells/well with and without test reagents, and after 24 hours, we added the XTT reagent. We allowed the reaction to proceed for 3 hours before measuring abundance of the reduced formazan derivative, as outlined by the manufacturer.

2.15 Statistics

Analysis were performed using SAS JMP Pro, versions 11.0 and 13.0, or GraphPad Prism, version 6.04. Data are shown as means±SD from one experiment, with the number of replicates provided in the text and figure legends. Each experiment was repeated at least three times using independent platings of cells. To compare two means, data were analyzed by two-way t-test; to compare more than two means, we used one-way ANOVA, followed by Tukey HSD test. For determining drug dose-effect parameters, the responses associated with each drug concentration were plotted vs. log drug concentration and the data were fitted to the equation Y=Bottom + (Top-Bottom)/(1+10^((LogIC50-X)*HillSlope)), where Y = response and X = log₁₀ (drug concentration), using PRISM 6.04 for calculating drug LogIC₅₀ values. To compare results between cell lines, or the results of different treatments for the same cell line, the data were normalized between 0 and 100% using the PRISM Normalize function, and the resulting data points were fit to the equation Y=100/(1+10^((LogIC50-X)*HillSlope)). Note, in the figures the parameter ‘LogXC₅₀’ is reported as ‘EC₅₀’ for the drug concentration that elicits 50% of the maximum response, and as ‘IC₅₀’ for the concentration inhibiting the response by 50%, as appropriate. In applying these procedures to oxygen uptake, the data points for each drug concentration were the respective slopes of the oxygen uptake curves.

3. RESULTS

3.1.1 TOM40 protein levels in brain

There was a trend toward higher Tom40 levels in brain tissue from APOE ε3/ε4 individuals who were poly-T S/L carriers compared with VL/L carriers, but the statistical significance of this result depended on the nature of the factor used for normalization. We conducted these measurements before we adopted our current practice of using the total measurable protein in each lane as a normalization standard, and we used two different cellular markers for normalization. Expression was ~20% greater, and the results were statistically significant, using the mitochondrial marker PDHE1α, (Figure 1A), but using the cytosolic marker α-tubulin, the difference was reduced to ~10% and was not significant (Figure 1B). Among APOE ε3/ε3 subjects, expression did not differ significantly between S/S or VL/VL homozygotes no matter the choice of normalization parameter (data not shown).

Figure 1. Brain TOM40 levels as a function of TOMM40 ‘523 genotype.

TOM40 levels trended higher in S/L vs. VL/L brain samples, but there was no difference in S/S vs. VL/VL samples. Brain tissue were homogenized, fractionated and probed for Tom40 as described in Materials and Methods.

3.1.2 TOM40 protein levels in overexpressing HeLa cells

We used a HeLa cell transfection system to determine how over expressing TOM40 affects cellular and mitochondrial function. One advantage of this approach is that it involves a uniform genetic background, simplifying interpretation of the results. For reference, we confirmed the genotypes of the cells with which these experiments were conducted are APOE ε3/4 and TOMM40 S/VL (T = 15/34). We evaluated five different transgenic HeLa cell lines: two clonal isolates resulting from transfection with a vector containing only the selection marker for G418 (HeLaC-1 and HeLaC-2 cells), and three clonal isolates resulting from transfection with a vector expressing a FLAG-tagged full-length human TOM40 protein (TOM40-1, -2, and -3 cells). We also used non-transfected HeLa cells obtained from ATCC via the Duke Cancer Center Cell Repository as controls for some experiments. Hereinafter, we refer to these cells as HeLa nulls, to distinguish them from the HeLaC controls. The levels of TOM40 in the control HeLaC-1 and HeLaC-2 cells represent endogenous levels of human TOM40 attained at 70 – 80% confluence under our growth conditions, and did not differ between the C2 and C3 lines (data not shown). Except where noted otherwise, we used the HeLaC-1 line as the control for comparison with the over-expressing lines in the subsequent experiments. TOM40 levels in TOM-1 and TOM-2 were 3 – 4 times as high as in HeLaC-1 cells (Figure 2), using HeLa null cells as the standard, and expression was stable in both lines through multiple thaws (not shown).

3.2 Subcellular distribution of TOM40

In addition to its widely recognized mitochondrial localization, ectopic localization of TOM40 on the cell surface has been reported [64]. For this reason, and because overexpression itself might cause abnormal localization of TOM40, we examined its subcellular localization in HeLa over-expressing cells. Immunohistochemical staining confirmed increased TOM40 expression in TOM40-2 cells compared with HeLaC-1 cells (Figure 3A) and in both cell types the fluorescent signal of the immunostained TOM40 overlapped almost completely with the MitoTracker Orange signal, confirming that over expression of TOM40 did not alter its subcellular distribution compared with its distribution in control cells. We confirmed these results using subcellular fractionation. Following differential centrifugation of homogenates from both HeLaC and TOM40 cells, we recovered most of the starting TOM40 in the heavy (11,000 × g) and light (16,000 × g) membrane fractions, and a small proportion in the 16,000 × g supernatant fraction. In each fraction, the proportion of starting TOM40 recovered was nearly the same for both the HeLaC-1 and TOM40-2 cells (Figure 3B).

Figure 3. Subcellular distribution of TOM40.

A. Immunocytochemical localization of TOM40. To observe the subcellular distribution of TOM40 in situ, cells were exposed to DAPI stain (a, blue) to label nuclei, MitoTracker Orange (b, red) to label mitochondria, and anti-TOM40 antibody (c, green), as described in Materials and Methods. The overlays show that TOM40 is expressed almost exclusively in mitochondria in both the control and the Tom40 over-expressing cells. A. HeLaC-1 control cells. B. TOM40-2 cells. B. Subcellular fractionation. Cells were grown in 145 mm dishes, and harvested and homogenized as described in Materials and Methods when the cultures had attained 70 – 80% confluence. Cell fractions were prepared by differential centrifugation. TOM 40 was determined by Western blotting of the total homogenate of both cell lines (not shown) and in each respective fraction, as described in Materials and Methods. All samples were fractionated on the same SDS-gel and developed on the same blot. The graph represents one experiment, which was repeated two additional times with similar results. HMF:”Heavy” mitochondrial pellet; S1: Post-HMF supernatant fraction; LMF: “Light” mitochondrial pellet; S2:Post-LMF supernatant fraction.

3.3 Mitochondrial protein abundance in TOM40 over-expressing and control cells

We reasoned that unregulated TOM40 expression might alter the overall mitochondrial composition, and to determine if this occurred, we quantified proteins from three distinct mitochondrial compartments, the outer membrane, the matrix, and the inner membrane. The levels of TOM20, a peptide receptor bound with TOM40 in the TOM complex, was also elevated in the TOM40 over-expressing cells (Figure 4A & B), as was expression of TOM22, another peptide receptor of the TOM complex (data not shown). The matrix proteins HSPA9/Mortalin (Figure 4A & B) and PDHE1α (Figure 4A & B) were also expressed at higher levels in TOM40 over-expressing cells, as were the activities of α-ketoglutarate dehydrogenase (α-KGDH), that also resides in the matrix, and complex I and complex IV of the oxidative phosphorylation system, that reside on the inner mitochondrial membrane (Figure 3C). During the course of these experiments, we noticed expression of TOM40 protein varies approximately inversely with the confluence of the growth cultures (data not shown). When we pooled the TOM40 and TOM20 expression results from these experiments, we found a linear relationship, with a slope of nearly one (Figure 4A). By contrast, while the expression of HSPA9 is also positively related to TOM40 expression there is considerable variation about the regression line and the slope is less than the slope for TOM20 (Figure 4B). While these data are purely observational, they suggest that the expression of members of the TOM complex may be co-regulated, but the relationship between expression of TOM40 and mitochondrial proteins that are not part of the TOM complex is more complex.

Figure 4. TOM40 overexpression is associated with increased levels of the TOM complex outer membrane protein TOM20, the matrix proteins HSPA9/GPR75/mortalin, PDHE1α and α-ketoglutarate dehydrogenase, and complex I and complex IV of the inner membrane.

Cells were grown to 75 – 80% confluence in 100 mm dishes, and extracted for Western blot analysis as described in Materials and Methods (A, B), or they were extracted in lauryl maltoside detergent solution (Abcam, C) and used for activity assays of a-KDGH, complex I and complex IV, using kits provided by Abcam, according to the manufacturer’s instructions. for (A) Western blot data. Serial dilutions were analyzed as described in Materials and Methods. HSPA9 (~75kD) and PDHE1α (~42kD) were fractionated in the same gel and blotted onto PVDF membrane. Tom20 was analyzed separately and blotted onto methylcellulose. (B) Quantification of the immunoblots. The data are means ± SD for four (Tom20) or 3 dilutions from each experiment shown. (C) Activities of αKDGH (C1), complex I (C2) or complex IV (C3). Both cell lines were analyzed simultaneously in the same gels A, B or same enzyme assay (C). Representative data are shown. Each experiment was repeated three times using cell extracts from independent cell passages, with similar results.

3.4. Mitochondrial abundance

The foregoing experiments suggest increased expression of TOM40 may lead to increased mitochondrial mass in over-expressing cells. MitoTracker Green fluorescence did not differ between the respective cell types (Figure 5A). Additional experiments with other control and TOM40 over-expressing clonal isolates led to the same result (data not shown). These results suggest the increased mitochondrial protein expression documented in section 3.3 reflect increased protein packing in individual mitochondria, leading to the question of whether this increased packing causes measureable mitochondrial stress. To address this question, we measured lipid staining of autolysosomes and pre-autolysosomes as a surrogate measure of mitophagy[65]. Staining was higher in both cell lines after 24-hour exposure to 10 μM FCCP than in cells exposed to DMSO alone, but there was no difference between the control and over expressing lines, either with or without FCCP (Figure 5B). This result suggests the increased mitochondrial content of proteins in the TOM40 cells does not alter the extent of mitophagy compared with that occurring in control cells.

Figure 5. The levels of TOM40 and TOM20 were positively correlated.

TOM40 and TOM20 were determined simultaneously in extracts of HeLaC-1, TOM40-1 and non-transfected control HeLa cells, that were grown in the presence of 25 or 5 mM glucose, and that were harvested from subconfluent or confluent cultures. Each point represents a single extract.

3.5 Doubling times

We determined cell-doubling times as another measure of whether over expression of TOM40 affects cellular homeostasis, either directly or indirectly through effects on mitochondrial function. Kim et al reported over expression of TOM40 reduced cell culture doubling times compared with matched controls[66]. By contrast, we found that over expression of TOM40 in HeLa cells did not alter growth rates. In the standardly used DMEM containing 25mM glucose, the mean doubling times for control HeLaC-1 cells were 0.9418 (95% CI, 0.8753 – 1.019, n=6), and 0.9345 (95% CI, 0.8624 – 1.020, n=6) for the TOM40-1 line. Reducing the glucose content of the DMEM to 5 mM increased the doubling times of both cell lines by about 20%. We conclude neither the over expression system per se, nor the increased abundance of TOM40 specifically, disrupted cell division regulation or mitochondrial function related to the supply of ATP or other substrates essential for cell division.

3.6. Bioenergetics

3.6.1. Mitochondrial membrane potential and cellular respiration

Basal cellular ATP levels (Figure 6A), mitochondrial membrane potential (Figure 6B, Supplemental Figure 1A) and oxygen utilization rates (Figure 6C) were higher in TOM40 cells than in the controls. To determine how over expression of TOM40 affects cellular respiratory control, we first established optimal concentrations of oligomycin, FCCP, rotenone and antimycin A in preliminary experiments. Regarding oligomycin, the ED₅₀’s and shapes of the dose-response effects of oligomycin on ATP levels and the mitochondrial membrane potential were indistinguishable in the two cell types. They reached a maximum at 1 μM, and were flat between 1 – 10 μM; we used 4 μM in the subsequent respiratory control experiments described below (data not shown).

Figure 6. MitoTracker Green staining (A) and autolysosome (B) detection.

(A).MitoTracker Green staining was the same in both the control and the TOM40 over-expressing cells. Ninety six-well plates were inoculated with 12×10³ cells/well. Twenty-four hours after plating, the growth medium was replaced with pyruvate- and malate-supplemented EBSS containing 150 nM MitoTracker green, and the fluorescence was determined after incubating for one hour at 37°C, as described in Materials and Methods. (B). Over-night exposure to low FCCP concentration elicited autophagy in both the HeLaC-1 controls and TOM40-1 over-expressing cells. There was a non-significant trend toward greater vacuole staining in the control cells. Cells were inoculated into 96-well plates and treated with FCCP as described in the Materials and Methods. For both (A) and (B), the data shown are means±StDev for n=8 wells from a single experiment and were analyzed by two-way t-tests.

3.6.2. FCCP

FCCP increased oxygen consumption and decreased the membrane potential with similar ED₅₀’s in both cell types. The shared logED₅₀ (±StDev) for lowering the membrane potential was −5.372±0.047 (r²=0.457, p=0.606, n=8); and was −5.473±0.052 (r²=0.572, p=0.711, n=8) for stimulating oxygen consumption. Supplemental Figure 1C, D also demonstrates the bi-phasic nature of the FCCP effect on oxygen utilization in both cell types, as previously pointed out for other cultured cells [67–69]. In each of five individual dose-response experiments, conducted using separate platings of cells, oxygen utilization reached a maximum at 10 μM FCCP for both cell types, which we used in all subsequent experiments with FCCP.

3.6.3 Rotenone and antimycin A

Rotenone inhibited oxygen utilization in both cell types, but HeLa control cells were more sensitive to rotenone-mediated effects than the TOM40 over expressing cells (Figure 7A; mean±STDEV logIC₅₀, HeLaC-1 and TOM40-1, respectively: −6.592±0.113, −7.340±0.79, P=0.0004, n=11). This may reflect the fact complex I activity is higher in the TOM40 cells than in the controls (Figure 5). Although rotenone suppressed cellular ATP levels with the same dose-response characteristics in both cell types (Figure 7B, shared logIC₅₀±STDEV, −7.686±0.148, P=0.1929, n=60), ATP levels remained higher in the over-expressing cells than in the controls throughout the dose-response range. In contrast to the effects of rotenone, the two cell types were equally sensitive to antimycin A-mediated suppression of oxygen consumption (Figure 8A). As was the situation for rotenone, the logED₅₀ for ATP lowering was the same in both cell types, and ATP levels were greater in the over-expressing cells than the controls at all antimycin A concentrations (Figure 8B). The relative insensitivity of ATP levels to both compounds likely reflects the importance of glycogenolysis-driven glycolysis to the maintenance of cellular ATP pools[70] during the time-course of these experiments. Note, as described in Materials and Methods, we routinely used 1μM antimycin A to suppress cellular oxygen consumption in our respiration experiments.

Figure 7. Basal ATP, mitochondrial membrane potential and oxygen consumption.

Basal levels of cellular ATP (A), respiration (B), and the mitochondrial membrane potential (C) were greater in the TOM40 over-expressing cells than in the controls. We seeded 96-well plates with 12×10³ cells/well, and 24 hours later, we replaced the growth medium with EBSS without or with the indicated FCCP concentration. Two hours later, we measured ATP and TMRM fluorescence (A, B). We initiated oxygen consumption readings in another set. We measured MitoXpress phosphorescence for two hours as described in Materials and Methods, and the data represent the linear portions of the oxygen consumption curves. For all three panels, the data shown are means±StDev for n=8 wells from single experiments and were analyzed by two-way t-tests (A) or one-way ANOVA followed by Tukey’s HSD test (B, C).

Figure 8. Rotenone effects on oxygen consumption, cellular ATP levels and ROS production.

HeLaC-1 cells are more sensitive than TOM40 overexpressing cells to inhibition of oxygen consumption by rotenone. (A) Respiration. Cells were incubated with 3 μM FCCP and increasing concentrations of rotenone. Readings were conducted every minute for two hours, and oxygen uptake was calculated as described in Materials and Methods. Each point represents maximum uptake rate for the respective rotenone concentration. For clarity, only the means for n=6 determinations are shown, but all the data were used for calculating the respective LogIC₅₀ and HillSlope values. The experiment was repeated three times, using a separate cell plating for each, with similar results. (B) ATP levels. The LogIC₅₀ for suppression of cellular ATP levels was the same in both cell types. (C) ROS levels. Basal ROS levels were the same in both cell types, and 1 μM FCCP elicited ROS synthesis in both, to approximately the same extent. The data shown for (C) are means±StDev for n=8 wells from single experiments and. The ROS data were analyzed by one -way ANOVA followed by Tukey’s HSD test.

We next determined how over-expression of TOM40 affected the generation of reactive oxygen species caused by inhibition of complex I and complex III [71–73], using the fluorescent dye 2′,7′-dihydrofluorscein. Throughout a series of side-by-side comparisons that involved different thaws of control and over expressing cells, endogenous ROS values did not differ significantly between the cell lines (data not shown). One μM rotenone, which was at the plateau of both the respiration and ATP dose-response curves, stimulated ROS production in both cell types, but the extent of increase did not differ between them (Figure 7C). Antimycin A also stimulated ROS production in both cell types with similar logED₅₀’s (Figure 8C, shared mean logED₅₀±SEM, −4.496±0.1475, P=0.2738, n=77). However, unlike the case for rotenone-elicited ROS production, the fold-increase was greater in the HeLaC-1 controls than in the TOM40-1 over expressing cells (mean difference, −67.89, 95% CI, −98.78 − −37, p=0.0004, n=12).

3.6.4 Respiratory control

To evaluate the effects of over-expressing TOM40 on respiratory control, we measured respiration without any additions (basal respiration) or in the presence of oligomycin (4 μM), plus or minus FCCP (10 μM) and in the absence or presence of 1 μM rotenone. Under basal conditions, mitochondrial respiration accounted for 95±2% of total respiration in HeLaC cells and 83±9% in TOM40 cells (means±STDEV, n=48, P=0.0001). The endogenous mitochondrial respiration accounted for 66±15% and 55±24%, respectively, of maximum FCCP-stimulated respiration (means±STDEV, n=30, NS). In HeLaC cells, proton leak equaled 47±12% of the endogenous mitochondrial respiration and 70±10% in TOM40-2 cells (n=30; P=0.0075). ATP synthesis accounted for 56±13% and 28±9%, respectively, of endogenous mitochondrial metabolism (n=30; P=0.0009). The spare respiratory capacity was higher in the TOM40 over-expressing cells than in the HeLaC controls (difference of means, HeLaC1 vs. TOM40-1, −1.748e-5, 95%CI (−2.176e-6, −3.278e-5), n=48; HeLaC-1 vs. TOM40-2, −2.010e-5, 95% CI (−2.446e-6, −3.775e-5), n=48). The respiratory control ratio also was higher in the TOM40 over-expressing cells than controls (HeLaC-1, 1.478±0.94 vs TOM40-1, 2.096±1.072, n=30, P=0.0102).

3.6.5. Effects of TOM40 over expression on cellular response to beta-amyloid

Next, we determined if over expression of TOM40 would protect mitochondrial function from damage elicited by beta-amyloid. Beta-amyloid has been found in association with mitochondria from normal and LOAD human brain samples, in brain mitochondria from animal models of familial AD, and with mitochondria in cell culture models, including yeast. [22–25, 74–78]. Mitochondrial beta-amyloid is correlated with mitochondrial dysfunction, including suppression of the mitochondrial membrane potential[79], and inhibition of respiratory complex I[80], complex IV[25], ABAD[24, 81], and SOD1[82]. Mitochondrially associated Aβ may be derived from extracellular pools of Aβ[31], or via cleavage of APP misdirected to mitochondria[83, 84] followed by direct internalization via the TOM complex[26]. On the other hand, evidence recently published by Cenini et al. suggest that mitochondrially-associated Aβ is not internalized but is instead non-specifically bound to the mitochondrial outer membrane[30]. Many of the adverse effects of Aβ on mitochondrial function may be due to inhibition of mitochondrial protein import via direct interactions of Aβ with mitochondrial precursor proteins[30], or with the preprotein processing machinery[85].

To determine the effects of over-expressing TOM40 on the consequences of exposure to Aβ, we measured the mitochondrial membrane potential and cell viability in cultures with and without added Aβ. In control experiments, we observed fluorescent label accumulated predominantly over mitochondria within 25 hours of addition of fluorescent-labeled beta-amyloid (Aβ_1–42 Hylite 488 in a solution of 5 μM Aβ) to the cultures (data not shown). As shown in figure 9A, 24-hour exposure of cultures to (unlabeled) Aβ_1–42 reduced the mitochondrial membrane potential, as reflected by the TMRM fluorescence, ca. 14% (compared with fluorescence in untreated cultures) in HeLaC-1 control cultures, but only by ca. 2% in cultures of TOM40-1 over-expressing cells (difference between means of differences, 2893.7, 95% CI (1486.3, 4303), n=27), and it lowered cellular ATP levels by 13% in control cells vs. 5% in TOM40-1 cells (Figure 9B, difference between means of differences, 848, 95% CI, 195.763, 1500.237), n=18). Exposure to Aβ also reduced cell viability by 19% in HeLaC-1 cultures but only by 2% in TOM40C-1 cells (Figure 9C, difference between means of differences, −0.17184, 95% CI, (−0.25183, −0.09185), n=54). In an independent trial, Aβ reduced viability 23% in HeLaC-1 cultures by but only by 14% in cultures of TOM40-2 cells (data not shown, difference between means of differences, −0.0947, 95% CI (−0.13887, −0.05053) n=27). On the other hand, over expression of TOM40 did not prevent Ab-elicited increases in cellular ROS (difference between means of differences, −68, 95% CI (−192.5, 56.5), n=18, data not shown). Therefore, over expression of TOM40 confers resistance to some, but not all, of the mitotoxic effects of internalized beta-amyloid peptide.

Figure 9. Antimycin A effects on oxygen consumption, cellular ATP levels, and ROS production.

Oxygen consumption (A) and cellular ATP levels (B) were equally sensitive to the effects of antimycin A in control and over-expressing cells. (C) ROS production was more sensitive to Antimycin A in the control cells than TOM40 over-expressing cells. Cells were incubated with 3 μM FCCP and increasing concentrations of Antimycin A. Oxygen consumption was measured as described in Materials and Methods. Each point represents maximum uptake rate for the respective antimycin A concentration. For clarity, only the means for n=6 determinations are shown, but all the data were used for calculating the respective LogIC₅₀ values. (B) Antimycin A suppresses cellular ATP levels; the IC₅₀ was the same for both cell types. (C) Antimycin A increased cellular ROS levels in HeLaC cells but not in TOM40 overexpressing cells. The reported LogEC₅₀ is for HeLaC cells.

4. DISCUSSION

Recently, our group has shown that specific TOMM40-APOE haplotypes are associated with cognitive decline in older individuals of European ancestry. Specifically, global cognition declined more rapidly in individuals who were homozygous for APOE ε3 and for the short poly-T in intron 6 of TOMM40 than in individuals who carried at least one copy of the VL-poly T[86]. Previously, we showed that TOMM40 mRNA expression was greater in brain samples from cognitively normal APOE ε3/ε3 individuals who were homozygous for the VL-poly T than in those who were homozygous for the short TOMM40 poly-T allele [46]. In the current study, we did not detect differences in the TOM40 protein levels in brain samples from these individuals, using Western blotting to measure TOM40. Given that the average correlation coefficient between mRNA and bespoke protein levels is only about 40%[87] [88] [89], this result may not be surprising, and may have have biological and technical causes. One technical challenge brain studies face is the choice of an appropriate normalization standard. We have adopted an approach of using the total protein measurement in a lane as the standard, as described in Materials and Methods, but our experiments with brain tissue took place before we implemented this approach, and we relied on ‘housekeeping’ proteins as normalization standards. In preliminary experiments we tested a number of cytosolic proteins, and none performed any differently than α-tubulin. Because it is likely that mitochondria are not distributed similarly to α-tubulin in the CNS, we also normalized brain TOM40 against PDHE1α because both share the same subcellular distribution. However, as shown above in Figure 3B, mitochondrial PDHE1α levels depend on TOM40 and therefore this choice also introduces a bias. An unbiased proteome-wide analysis may provide more accurate information, and we have developed a mass spectrometric assay for TOM40 (unpublished results) that we will use in conjunction with such an approach in future studies.

Because TOMM40 encodes the mitochondrial protein import channel, these results led us to investigate how increased TOMM40 expression affects mitochondrial function. In this work, we used a transgenic HeLa system engineered to overexpress TOM40 protein to begin to address this question. Earlier studies documented the lethality of homozygous TOM40 deletions in yeast[49], fungi[50] and mice[51], and the bioenergetic and neurologic phenotypes of the hemizygous knock-down mouse[51]. The effects of structural mutations on TOM40 function also have been described for yeast and fungi [50, 90–94]. However, over-expression of TOM40 has not been studied widely.

We discovered that although the abundance of mitochondria was the same in cells over-expressing TOM40 as in controls transfected with a vector that lacked the TOM40 coding sequence, the over-expressing cells maintained higher mitochondrial membrane potentials, experienced higher endogenous respiration rates, and had greater spare respiratory capacities and maintained higher endogenous ATP levels than the controls. Moreover, TOM40 over-expressing cells resisted the cytotoxic effects of Aβ_1–42 added to the medium, and continued to maintain higher mitochondrial membrane potentials and cellular ATP levels in the presence of Aβ than the control cells.

The mitochondrial bioenergetic profile of the TOM40 over-expressing cells vis a vis that of control cells likely results from enhanced uptake and incorporation of mitochondrial proteins. Over-expressing HeLa clones produced two-to-four times as much TOM40 protein as the controls. The subcellular distribution of overexpressed TOM40 was the same as that as endogenous TOM40, and together these data suggest that overexpression did not disrupt trafficking or mitochondrial insertion of the TOM40 precursor protein, or mitochondrial biogenesis. We determined the protein expression levels of proteins targeted to three mitochondrial compartments to learn if TOM40 overexpression had selective effects on the uptake and incorporation of specific classes of mitochondrial proteins. TOM40 overexpression was associated with elevated levels of the outer membrane proteins TOM20 and TOM22; but not of anion channel VDAC. Possibly, members of the TOM complex are co-regulated, similar to the regulation of genes encoding components of the oxidative phosphorylation complexes[95]. Because TOM40 is incorporated into mitochondria via pre-existing TOM complexes [96–98], increased abundance of the TOM40 protein might be the trigger in the over-expressing cells that elicits expression of other members of the complex. The matrix protein HSPA9 functions, in part, in cooperation with the mitochondrial protein import machinery by promoting folding of nascent imported proteins[99]. Its expression was higher in TOM40 over-expressing cells than in controls, although its expression was not as highly correlated with TOM40 expression as was the expression of TOM20. This may not be surprising, however, because HSPA9 is not a bespoke TOM complex member or mitochondrial protein, and also functions at the plasma membrane, the ER and a variety of cytosolic vesicles [100–103], and in the cytosol[104]. None-the-less, increased expression of HSPA9 in the TOM40 over-expressing cells is consistent with enhanced biogenesis of mitochondrial proteins, and it may contribute the enhanced bioenergetic profile in TOM40 over-expressing cells, by preventing mis-folding of mitochondrial proteins associated with pathogenic factors associated with AD and other stressors[85, 105].

TOM40 over-expression was also associated with increased PDHE1α and α-KGDH, and complexes I and IV of the electron transport system.α-KGDH and the two oxidative phosphorylation complexes are flux-controlling steps in the TCA cycle and oxidative phosphorylation, respectively [68, 106–111], and PDH is rate-controlling for acetyl CoA production, which feeds into bioenergetics, fatty acid and lipid synthesis and neurotransmitter synthesis. Increased TOM40 expression might be protective for the expression of these complexes. The enhanced expression/activities of these complexes, and possibly of other members of the TCA cycle and oxidative phosphorylation pathways we did not measure, likely contribute to the robust mitochondrial bioenergetic profile of the TOM40 over-expressing cells.

Previously, Hedskog et al. reported there were no measurable effects of the TOMM40 poly-T polymorphism on mitochondrial function or morphology in primary fibroblasts from cognitively normal subjects who were APOE ε3/ε4 and either S/L or VL/L with respect to the ‘523’ poly-T length polymorphism[48]. We[46] and others[45, 112] showed the poly-T is regulatory and determines the level of TOMM40 mRNA expression. In brain samples from APOE ε3/ε3 individuals, TOMM40 mRNA expression increased dose-dependently with the number of VL alleles, and expression of a luciferase reporter gene was higher in a construct incorporating the VL poly-T length than the S length[46]. Payton et al showed the S/S poly-T length acted as a repressor of luciferase expression in a reporter construct[45]. By contrast, TOM40 expression levels did not differ between the S/L and VL/L populations in the Hedskog et al study. Because of this, they were not able to study the effects of altered TOMM40 mRNA or protein on mitochondrial function. In addition, because all of their study subjects were APOE ε3/ε4 heterozygotes, their results may have reflected contributions of the APOE ε4 allele to the mitochondrial phenotype[18, 54]. Although HeLa cells are also APOE ε3/ε4, HeLa cells do not express ApoE protein[53], and the exogenous expression system we employed obviated possible transcriptional effects of the endogenous genomic APOE gene on TOM40 expression driven from the exogenous plasmid. In addition, the control and over-expressing cultures we used were isogenic, reducing concerns that gene differences outside the TOMM40-APOE region might have affected our results.

One disadvantage of our system is that expression from the vector is independent of the normal regulatory control for TOMM40 expression. Because of this, we cannot ignore the possibility that our results were influenced by stoichiometric imbalances between TOM40 and its usual binding partners in the TOM and TIM complexes, or promiscuous interactions of TOM40 with atypical binding partners. Other disadvantages of the HeLa system are that the metabolic and bioenergetic properties of transformed cells differ from those of non-transformed cells, and that HeLa are derived from tissue distinct from those found in the CNS. For these reasons, the responses of HeLa cells to TOM40 over expression may not accurately predict the response of normal brain tissue. To overcome these difficulties we are developing targeted replacement mouse models, in which we have replaced the endogenous mouse Tomm40-Apoe region with the homologous human TOMM40-APOE genomic region. We are creating models of each of the three predominant poly-T lengths (S, L and VL), and these models will facilitate studies of the effects of the TOMM40 poly-T at the cell, organ and whole animal levels.

Variations in TOMM40 protein and mRNA levels are associated with a number of disease states, and the variation in neurodegeneration and cancer are especially noteworthy. TOMM40 expression is lower in blood from human subjects with Alzheimer’s disease than in the blood from aged-matched controls [113–115]. TOM40 protein levels also were reduced in neurons isolated by laser capture microdissection from a mouse PD model compared with levels in neurons from unaffected controls[116]. By contrast, TOMM40 mRNA and protein levels were elevated in tumor cells [66] compared with expression levels in adjacent, non-disease tissue. Learning the biological consequences of altered TOM40 expression will contribute to a deeper understanding the role TOM40 plays in regulating mitochondrial function, and possible new insights to the pathogenesis of these diseases..

Figure 10. Over-expression of TOM40 protected cells from mitotoxic and cytotoxic effects of exogenous Aβ.

Aβ_1–42was prepared as described in Materials and Methods and applied to cells in 96-well plates (5μM, final concentration). After 24 hours TMRM (A) or XTT (C) were added to the medium, and incubations continued for two (A, B) or four (C) hours. At the designated times, TMRM fluorescence and XTT fluorescence were measured, or cells were extracted for ATP measurements, as described in Materials and Methods. Aβ lowered the mitochondrial membrane potential (A), and cellular ATP levels (B) and reduced cell viability (C) in the control HeLaC-1 cells but not the TOM40-1 cells. The data are means±SD of 8 replicates in a single experiment, and the data were analyzed by one-way ANOVA followed by Tukey’s HSD test.

HIGHLIGHTS.

• A variable-length poly-T variant in intron 6 of the TOMM40 gene, rs10524523, is associated with risk and age-of-onset of late-onset Alzheimer’s disease. The alleles at this locus are classified as Short (S), Long (L) or Very long (VL).

• S/VL and VL/VL genotypes are more protective than S/S.

• The VL poly-T results in higher expression than the S poly-T in luciferase expression systems.

• Over-expression of TOM40 in HeLa cells enhances mitochondrial efficiency.

• TOM40 over-expression is protective against beta-amyloid-induced cellular damage