Mitochondrial bioenergetic deficit precedes Alzheimer\'s pathology in female mouse model of Alzheimer\'s disease

Mitochondrial bioenergetic deficit precedes Alzheimer’s pathology in female mouse model of Alzheimer’s disease Jia Yao, Ronald W. Irwin, Liqin Zhao, Jon Nilsen1, Ryan T. Hamilton, and Roberta Diaz Brinton2 Department of Pharmacology and Pharmaceutical Sciences and Program in Neuroscience, Pharmaceutical Sciences Center, University of Southern California, Los Angeles, CA 90033 Edited by Bruce S. McEwen, The Rockefeller University, New York, NY, and approved June 29, 2009 (received for review April 2, 2009)

Mitochondrial dysfunction has been proposed to play a pivotal role in neurodegenerative diseases, including Alzheimer’s disease (AD). To address whether mitochondrial dysfunction precedes the development of AD pathology, we conducted mitochondrial functional analyses in female triple transgenic Alzheimer’s mice (3xTg-AD) and age-matched nontransgenic (nonTg). Mitochondrial dysfunction in the 3xTg-AD brain was evidenced by decreased mitochondrial respiration and decreased pyruvate dehydrogenase (PDH) protein level and activity as early as 3 months of age. 3xTg-AD mice also exhibited increased oxidative stress as manifested by increased hydrogen peroxide production and lipid peroxidation. Mitochondrial amyloid beta (A␤) level in the 3xTg-AD mice was significantly increased at 9 months and temporally correlated with increased level of A␤ binding to alcohol dehydrogenase (ABAD). Embryonic neurons derived from 3xTg-AD mouse hippocampus exhibited significantly decreased mitochondrial respiration and increased glycolysis. Results of these analyses indicate that compromised mitochondrial function is evident in embryonic hippocampal neurons, continues unabated in females throughout the reproductive period, and is exacerbated during reproductive senescence. In nontransgenic control mice, oxidative stress was coincident with reproductive senescence and accompanied by a significant decline in mitochondrial function. Reproductive senescence in the 3xTg-AD mouse brain markedly exacerbated mitochondrial dysfunction. Collectively, the data indicate significant mitochondrial dysfunction occurs early in AD pathogenesis in a female AD mouse model. Mitochondrial dysfunction provides a plausible mechanistic rationale for the hypometabolism in brain that precedes AD diagnosis and suggests therapeutic targets for prevention of AD. ABAD 兩 aging 兩 bioenergetics 兩 brain hypometabolism 兩 mitochondria

T

he essential role of mitochondria in cellular bioenergetics and survival has been well established (1–3). Previous studies have suggested that mitochondrial dysfunction plays a central role in the pathogenesis of neurodegenerative disorders, including Alzheimer’s disease (AD) (1, 4). Alzheimer’s pathology is accompanied by a decrease in expression and activity of enzymes involved in mitochondrial bioenergetics, which would be expected to lead to compromised electron transport chain complex activity and reduced ATP synthesis (5). Further, in AD there is a generalized shift from glycolytic energy production toward use of an alternative fuel, ketone bodies. This is evidenced by a 45% reduction in cerebral glucose utilization in AD patients (6), which is paralleled by decrease in the expression of glycolytic enzymes coupled to a decrease in the activity of the pyruvate dehydrogenase (PDH) complex (5). Patients with incipient AD exhibit a utilization ratio of 2:1 glucose to alternative fuel, whereas comparably aged controls exhibit a ratio of 29:1, whereas young controls exclusively use glucose as with a ratio of 100:0 ratio (7). In addition to the lowered mitochondrial bioenergetic capacity, impairment of oxidative phosphorylation is associated with increased free radical production and the resultant oxidative damage. Overproduction of reactive oxygen species and higher oxidative stress is characteristic of brains from AD (8). Increased oxidative stress, coupled with dysregulation of 14670 –14675 兩 PNAS 兩 August 25, 2009 兩 vol. 106 兩 no. 34

calcium homeostasis and resulting apoptosis of vulnerable neuronal populations, are proposed to underlie the loss of synaptic activity and associated cognitive decline (9). In addition to the mitochondrial dysfunction in the clinically confirmed AD cases, multiple analyses have demonstrated that mitochondrial dysfunction is a plausible contributing factor in the pathogenesis of sporadic AD. The ‘‘cybrid model’’ of AD has provided evidence for mitochondrial dysfunction in AD pathogenesis (10). AD cybrid cells exhibit decreased COX activity, decreased mitochondrial membrane potential, decreased mitochondrial mobility and motility, increased oxidative stress, over activation of caspase-3, and increased A␤ production (11–13). Further, increased risk of AD occurs in offspring of women with AD, suggesting human maternal mitochondrial inheritance (14, 15). Collectively, these observations suggest a role for mitochondrial dysfunction in the pathogenesis of sporadic AD. To determine whether mitochondrial bioenergetic mechanisms are associated with AD pathogenesis, we assessed mitochondrial function in the triple transgenic AD mouse model (3xTg-AD) developed by Oddo, LaFerla, and colleagues (16, 17). This transgenic mouse strain bears mutations in three genes (human APPSWE, TauP301L, and PS1M146V genes) linked to AD and frontotemporal dementia (FTD) and exhibits an age-related neuropathological phenotype, including both amyloid beta (A␤) deposition and tau hyperphosphorylation (16, 18). To characterize the change in mitochondrial functions in this model, we conducted biochemical and functional assays on whole brain mitochondria isolated from both 3xTg-AD and nontransgenic (nonTg) female mice at different ages. The results presented herein indicate that mitochondrial dysfunction, especially that leading to compromised energy production, precedes plaque formation, the hallmark histopathology of AD. Collectively, current clinical findings and our data are suggestive of a potential causal role of mitochondrial dysfunction in AD pathogenesis. From a therapeutic perspective, these data support a strategy that targets mitochondrial bioenergetics to prevent or delay the development of AD. Results Age-Dependent Development of AD-Like Pathology in Female 3xTg-AD Mice. The triple-transgenic AD mouse model has been

demonstrated to exhibit an age-related neuropathological progression pattern (16, 18). To determine the temporal correlation Author contributions: J.Y., J.N., and R.D.B. designed research; J.Y., R.W.I., and R.T.H. performed research; L.Z. contributed new reagents/analytic tools; J.Y., R.W.I., J.N., R.T.H., and R.D.B. analyzed data; and J.Y., J.N., and R.D.B. wrote the paper. The authors declare no conflict of interest. This article is a PNAS Direct Submission. 1Present

address: Amgen, Inc., 1 Amgen Center Drive, Thousand Oaks, CA 91320.

2To

whom correspondence should be addressed at: Pharmacology and Pharmaceutical Sciences, Pharmaceutical Sciences Center, University of Southern California, 1985 Zonal Avenue PSC502, Los Angeles, CA 90089. E-mail: [email protected].

This article contains supporting information online at www.pnas.org/cgi/content/full/ 0903563106/DCSupplemental.

www.pnas.org兾cgi兾doi兾10.1073兾pnas.0903563106

Fig. 1. Age-related increase in A␤-IR in female 3xTg-AD mice. Representative images show A␤-IR in hippocampus CA1 region from female 3xTg-AD mice at 3, 6, 9, and 12 months and female nonTg mice at 12 months (Scale bar, 100 ␮m).

Decreased Expression and Activity of Key Regulatory Enzymes of Oxidative Phosphorylation in Female 3xTg-AD Mice. Decreased ex-

pression and activity of cytochrome c oxidase (COX) and PDH have been observed in postmortem brain tissue derived from Alzheimer’s patients (5). To determine if 3xTg-AD mice recapitulated these mitochondrial defects and to identify the temporal correlation with AD histopathology, hippocampal proteins were isolated from another set of intact female 3xTg-AD and nonTg mice at 3, 6, 9, and 12 months of age. Expression of PDH (PDH E1␣) and COX (COX subunit IV) was assessed by western blot analysis. PDH E1␣ expression in 3xTg-AD mice was decreased relative to agematched nonTg mice (Fig. 2A; P ⬍ 0.05, n ⫽ 6). Significantly decreased PDH E1␣ expression was evident as early as 3 months of age and was greatest at 12 months of age (Fig. 2 A). Likewise, COX IV expression was decreased in 3xTg-AD mice compared with age-matched nonTg mice and was significantly decreased at 9 months of age (Fig. 2B). To confirm that the changes in protein expression were indicative of changes in enzyme activity, PDH and COX activities were assessed in mitochondria isolated from whole forebrain of the same set of 3xTg-AD and nonTg mice used for the hippocampal PDH E1␣ and COX IV expression. Both PDH and COX activities were decreased in the aging 3xTg-AD mice as compared with the age-matched nonTg mice (Table 1). Whereas PDH E1␣ expression was significantly decreased as early as 3 months of age in the 3xTg-AD mice, significant decline in PDH activity was first evident at 9 months of age. The preserved PDH enzyme activity relative to the PDH E1␣ subunit expression in the 3- and 6-month-old 3xTg-AD mice may be indicative of compensatory up-regulation of

PDH activity by posttranslational modification. As with PDH activity, the decline in COX activity was also significantly decreased at 9 months of age (Table 1). Increased Oxidative Stress in Brain Mitochondria of 3xTg-AD Mice.

Mitochondrial dysfunction is associated with oxidative stress and development of AD neuropathology. To determine the oxidative load of brain mitochondria in the 3xTg-AD mice, we assessed the rate of hydrogen peroxide production and magnitude of lipid peroxidation in mitochondria isolated from whole forebrain of female 3xTg-AD and nonTg mice at 3, 6, 9, and 12 months of age. The Amplex Red hydrogen peroxide assay was used to determine the rates of hydrogen peroxide production of mitochondria in both state 4 (in the absence of ADP) and state 3 (in the presence of ADP) respiration. There was an age-associated increase in the rate of state 4 hydrogen peroxide production in both the 3xTg-AD and nonTg mice (Fig. 3; P ⬍ 0.05, n ⫽ 6). More importantly, there was a significant increase in the rate of state 4 hydrogen peroxide production in the 3xTg-AD mice as compared with age-matched nonTg mice, which was evident as early as 3 months of age and was most pronounced at 12 months of age (Fig. 3; P ⬍ 0.05, n ⫽ 6). Production of hydrogen peroxide in the nonTg female mice increased at 9 months of age to a level comparable to, and thus not significantly different from, that generated by comparable aged 3xTg-AD mice. Increased hydrogen peroxide production would be expected to lead to a rise in oxidative damage to cellular components. Therefore, we measured lipid peroxidation as an indicator of overall oxidative stress. Both whole forebrain mitochondria and hippocampal lysates were used to determine lipid peroxidation, which yielded comparable results. Correlated with the increased rate of hydrogen peroxide production, there was significant age-related increase in lipid peroxidation in both the 3xTg-AD and nonTg mice in both isolated mitochondria (Fig. 4; P ⬍ 0.05, n ⫽ 6) and hippocampal lysates. Further, there was a significant increase in lipid peroxidation in the 3xTg-AD mice as compared with age-matched nonTg mice in both isolated mitochondria (Fig. 4; P ⬍ 0.05, n ⫽ 6) and hippocampal lysates. Decreased Brain Mitochondrial Respiratory Efficiency in 3xTg-AD Mice. Decreased expression and activity of the key mitochondrial

regulatory enzymes, PDH and COX, would be expected to result

Table 1. Decreased PDH and COX activity in female 3xTg-AD mice Age, months 3 6 9 12

PDH activity (nmol/min/mg protein) nonTg 116.45 ⫾ 9.89 83.73 ⫾ 5.37 107.07 ⫾ 6.14 79.43 ⫾ 5.66

3xTg-AD 126.2 ⫾ 5.46 76.91 ⫾ 5.12 86.94 ⫾ 4.54* 56.59 ⫾ 1.37*

Relative COX activity (normalized to 3m nonTg) nonTg 100 ⫾ 7.25 94.81 ⫾ 7.48 100.2 ⫾ 11.62 69.1 ⫾ 2.60

3xTg-AD 105.8 ⫾ 6.07 92.22 ⫾ 7.33 74.65 ⫾ 9.48* 52.7 ⫾ 0.63*

Mitochondrial enzyme function in the aging female nonTg and 3xTg-AD mouse brain. Brain mitochondria isolated from both nonTg and 3xTg-AD mice were assessed for PDH and COX activity. Relative COX Activity was presented as the relative value normalized to that of 3 month nonTg female mice. Mean ⫾ SEM (*, P ⬍ 0.05 compared with age-matched nonTg group, n ⫽ 6). Yao et al.

PNAS 兩 August 25, 2009 兩 vol. 106 兩 no. 34 兩 14671

NEUROSCIENCE

between mitochondrial dysfunction and AD histopathology, we characterized the age-dependent development of amyloid pathology in the triple-transgenic mouse model. A␤ accumulation in the hippocampal CA1 region was assessed in both gonadally intact 3xTg-AD and nonTg female mice at 3, 6, 9, and 12 months of age. We observed minimal A␤-immunoreactivity (A␤-IR) in the hippocampal CA1 region in the 3xTg-AD mice at 3 months. A␤-IR increases in an age-dependent way in 3xTg-AD mice and at 12 months, overt amyloid plaque formation is observed, whereas no A␤-IR is observed in nonTg mice at all age groups (Fig. 1).

Fig. 4. 3xTg-AD female mice exhibit higher lipid peroxidation than agematched nonTg female mice. Whole brain mitochondria were isolated from intact female mice, and lipid peroxide levels were determined by the leucomethylene blue assay. Bars represent mean ⫾ SEM (*, P ⬍ 0.05 compared with age-matched nonTg group; n ⫽ 6).

Fig. 2. 3xTg-AD female mice have decreased PDH E1␣ and COX IV protein levels relative to age-matched nonTg female mice. Equal amount of hippocampal (for PDH E1␣) and mitochondrial (for COXIV) samples from both 3xTg-AD and nonTg mice of different age groups were loaded onto the gel. Expression of (A) PDH E1␣ subunit and (B) COX IV subunit were determined by western blot analysis. Bars represent mean relative expression ⫾ SEM (*, P ⬍ 0.05 compared with agematched nonTg group; n ⫽ 6).

in decreased oxidative phosphorylation and impaired mitochondrial respiratory efficiency. To determine if oxidative phosphorylation was altered in the 3xTg-AD mice, mitochondrial respi-

Fig. 3. 3xTg-AD female mice exhibit higher oxidative stress than age-matched nonTg female mice. Whole brain mitochondria were isolated from intact female mice, and rates of hydrogen peroxide production were determined in state 4 respiration by Amplex-Red hydrogen peroxide assay. Bars represent mean hydrogen peroxide production rates ⫾ SEM (*, P ⬍ 0.05 compared with agematched nonTg group; #, P ⬍ 0.05 compared between different age group within the same genotype; n ⫽ 6). 14672 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0903563106

ration was determined in freshly isolated whole forebrain mitochondria from female 3xTg-AD and nonTg mice at 3, 6, 9, and 12 months of age. Respiratory rate of isolated whole brain mitochondria was first determined using glutamate (5 mM) and malate (5 mM) as respiratory substrates. ADP addition to the mitochondrial suspension initiated state 3 respiration. Addition of the adenine nucleotide transporter inhibitor atractyloside reduced the rate of O2 consumption to that of state 40 respiration, limited by proton permeability of the inner membrane. In nonTg group, an age-related decline in the respiratory control ratio (RCR; state 3:state 40) was apparent from 3 to 9 months and reached statistical significance at 12 months when compared with 3 months. Similarly in 3xTg-AD mice, there was also an age-related decline in RCR, which also reached statistical significance at 12 months when compared with either 3, 6, or 9 months (Fig. 5A; P ⬍ 0.05, n ⫽ 6). More importantly, compared with the age-matched nonTg group, 3xTg-AD mice showed decreased RCR at each age, and this genotype-related impairment of mitochondrial respiration deteriorated with age and was most pronounced at 12 months of age (Fig. 5A; P ⬍ 0.05, n ⫽ 6). The efficiency of mitochondria can also be assessed by determining the rate of free radical leak, or the percent electron flow that reduces oxygen to ROS instead of reducing O2 to water by COX enzyme. Similar to the change in RCR, in both 3xTg-AD and nonTg mice, the age-related increase in the free radical leak with malate/ glutamate plus ADP (state 3) was apparent and reached statistical significance at 12 months when compared with either 3, 6, or 9 months (Fig. 5B; P ⬍ 0.05, n ⫽ 6). More importantly, the free radical leak was significantly increased in the 3xTg-AD mice as compared with age-matched nonTg mice at 3, 6, and 12 months of age (Fig. 5B; P ⬍ 0.05, n ⫽ 6). As with hydrogen peroxide generation, the lack of statistically significant effect between nonTg and 3xTg-AD females at 9 months of age was because of the rise in the free radical leak of the nonTg to a level comparable to that of the 3xTg-AD mice (Fig. 5B). To determine the cellular contribution to the mitochondrial deficits of 3xTg-AD mouse brain, basal cellular respiration and glycolysis in primary neuronal cultures from 3xTg-AD and nonTg mice were harvested on embryonic day 14 and cultured for 1 week in vitro. Oxygen consumption rates (OCR) and extracellular acidification rates (ECAR) were determined using the Seahorse XF-24 metabolic flux analyzer. In primary neuronal cultures ⬎80% of the OCR measured was because of oxidative phosphorylation (Fig. 6A), and ⬎89% of the ECAR measured was because of lactic acid production via glycolysis. Neurons derived from the 3xTg-AD mice exhibited significantly lower OCR relative to the OCR of nonTg neurons (Fig. 6A; P ⬍ 0.05, n ⫽ 3). Correlated with the decreased oxygen consumption was increased glycolysis, as evidenced by a significant increase in the ECAR (Fig. 6C). The addition of the ATP synthase inhibitor oligomycin (1 ␮M) resulted in ⬃60% decrease in OCR in neurons from both 3xTg-AD and nonTg mice Yao et al.

(Fig. 6), indicating that the measured oxygen consumption was largely driven by oxidative phosphorylation-coupled ATP generation. That the decrease in OCR in response to oligomycin was similar in both nonTg and 3xTg-AD groups indicates that the decline in oxygen consumption was not because of a direct impairment of ATP synthesis. The decrease in oxygen consumption in response to oligomycin was correlated with an increase in ECAR (Fig. 6C), indicating a shift to ATP production through glycolysis

via the Pasteur effect (19). The addition of a mitochondrial uncoupler (FCCP; 1 ␮M) resulted in a dramatic increase in OCR, as expected, giving an estimation of the maximal respiratory capacity of the mitochondria. Both the direct measurement of OCR as well as the percent increase over baseline in response to FCCP were significantly lower in hippocampal neurons derived from the 3xTg-AD mice as compared with the nonTg mice (Fig. 6 A and B; P ⬍ 0.05, n ⫽ 3). These data suggest an impairment of the reserve respiratory capacity in the 3xTg-AD neurons that would potentiate mitochondrial dysfunction in the face of increasing metabolic demand. Further, the decreased maximal respiratory capacity is consistent with the impaired COX activity in the 3xTg-AD mice. The addition of the complex I inhibitor rotenone resulted in a further reduction in OCR values to approximately 15% of baseline in neurons from both 3xTg-AD and nonTg mice (Fig. 5). The ⬃25% difference in OCR values between oligomycin and rotenone exposure in both 3xTg-AD and nonTg neurons indicates that both groups had equivalent oxygen consumption because of proton leakage. The residual OCR capacity in the presence of rotenone most likely represents cellular oxygen consumption by nonmitochondrial pathways. Increased Mitochondrial A␤ Levels in 3xTg-AD Mice. Previous studies indicated that A␤ interacts with the mitochondrial protein, A␤binding alcohol dehydrogenase (ABAD), and that binding of A␤ contributes to mitochondrial dysfunction (20, 21). To determine the relationship between mitochondrial A␤ and the above observed mitochondrial dysfunction, mitochondrial and hippocampal lysates from female 3xTg-AD and nonTg female mice at 3, 6, 9, and 12 months of age were analyzed by western blot for ABAD and mitochondrial A␤ levels. ABAD protein levels decreased with age in nonTg mice, whereas ABAD increased with age in 3xTg-AD mice, which was clearly apparent and significant by 9 and 12 months of age in the 3xTg-AD (Fig. 7A; P ⬍ 0.05, n ⫽ 6). Likewise, at 9 months, there was a significantly higher level of A␤ oligomer level in the mitochondria of 3xTg-AD mice as compared with agematched nonTg mice (Fig. 7B; P ⬍ 0.05, n ⫽ 6).

Fig. 6. Hippocampal neurons derived from 3xTg-AD mouse brain exhibit decreased mitochondrial respiration and increased glycolysis. Primary embryonic neurons derived from both 3xTg-AD and nonTg mice were cultured in Neurobasal medium plus B27 supplement for 7 days. OCR and ECAR were determined using Seahorse XF-24 Metabolic Flux analyzer. (A) OCRs in primary neurons from 3xTg-AD mice (black) have lower basal rates of mitochondrial respiration than primary neurons derived from nonTg mice (gray). Vertical lines indicate time of addition of mitochondrial inhibitors (A) oligomycin (1 ␮M), (B) FCCP (1 ␮M), or (C) rotenone (1 ␮M). The maximal respiratory capacity (FCCP) is significantly lower in neurons from 3xTg-AD mice than those from nonTg mice. (B) Percent change in mitochondrial respiration in response to mitochondrial inhibitors. Bars represent the mean change in OCR from baseline ⫾ SEM (*, P ⬍ 0.05 as compared with age-matched nonTg group; n ⫽ 3). (C) Temporal bioenergetic profiling (ECAR vs. OCR) of primary hippocampal neurons after exposure to mitochondrial inhibitors (A) oligomycin, (B) FCCP, and (C) rotenone (*, P ⬍ 0.05 as compared with nonTg cultures; n ⫽ 3).

Yao et al.

Discussion Increasing evidence implicates mitochondrial dysfunction in multiple neurodegenerative disorders (22). In this report, we demonstrated that the female 3xTg-AD mouse brain recapitulates multiple indicators of mitochondrial dysfunction found in the human AD patients, including decreased mitochondrial bioenergetics, increased oxidative stress, and increased mitochondrial amyloid load in the 3xTg-AD mouse model (4, 23). Moreover, mitochondrial dysfunction evident in embryonic neurons was sustained throughout postnatal and reproductive ages and was most apparent after reproductive senescence at 12 months of age. Of particular importance is that the onset of mitochondrial dysfunction in the 3xTg-AD female mouse precedes the onset of plaque formation, the classical histopathological marker of AD pathology. Although dysfunction of each of the individual components was not observed at the same PNAS 兩 August 25, 2009 兩 vol. 106 兩 no. 34 兩 14673

NEUROSCIENCE

Fig. 5. 3xTg-AD female mice exhibit decreased mitochondrial respiration relative to age-matched nonTg female mice. Whole brain mitochondria were isolated from intact female mice, and state 3:state 4 respiration was determined. Oxygen electrode measurements of respiration using isolated brain mitochondria from 3xTg-AD and nonTg mice were conducted in the presence of 5 mM L-malate, 5 mM L-glutamate, 410 ␮M ADP to initiate state 3 respiration, and atractyloside to induce state 40 respiration. Bars represent the mean ⫾ SEM of the RCR (state3:state 4 respiration) (*, P ⬍ 0.05 as compared with age-matched nonTg group; #, P ⬍ 0.05 compared between different age group within the same genotype; n ⫽ 6).

Fig. 7. Female 3xTg-AD mice have increased ABAD and mitochondrial A␤ protein levels relative to age-matched nonTg female mice. Equal amount of hippocampal (for ABAD) and mitochondrial (for A␤) samples from both 3xTg-AD and nonTg mice of different age groups were loaded onto the gel. Expression of ABAD and 16 kDa mitochondrial A␤ oligomer were determined by western blot analysis. (A) 3xTg-AD mice have increased ABAD level than age-matched nonTg mice; (B) increased mitochondrial 16 kDa A␤ oligomer in the 3xTg-AD mice at 9 months. Bars represent mean relative expression ⫾ SEM (*, P ⬍ 0.05 compared with age-matched nonTg group; n ⫽ 6).

time, mitochondrial dysfunction is more than the sum of its components. Sustained disturbances in the pathway balances or functional efficiency can lead to systems-level defects not observable in additional individual components until later. Such early imbalances can lead to accumulated changes in the mitochondria that later emerge as observable changes in other aspects of the system. Collectively, our findings indicate a critical link between the transgenes APPswe, PS1M146V, and TauP301L and mitochondrial dysfunction during early neuropathogenesis. Both the expression and activity of PDH and COX in 3xTg-AD mitochondria were significantly decreased. Reduced mitochondrial efficiency occurred in neurons as evidenced by the shift from oxidative phosphorylation to lactic acid producing glycolysis in primary neurons from 3xTg-AD mice. As oxidative phosphorylation is driven by the activity of terminal enzyme COX and PDH serves as the regulatory switch coupling glucose utilization to oxidative phosphorylation, we propose that in 3xTg-AD mice, deficits in the activity of these 2 enzymes reduces the substrate input and driving force for oxidative phosphorylation, resulting in increased ATP demand via other pathways. This unbalanced metabolic state coupled with the oxidative stress because of reduced ETC efficiency leads to impaired bioenergetics that impairs neuronal function and exacerbates neurodegeneration. The occurrence of these metabolic impairments before plaque formation recapitulates findings derived from Alzheimer brain tissue and indicates that mitochondrial dysfunction is an important early factor in the development of AD-like pathology. The concurrent increase in glycolysis in 3xTg-AD neurons could be a compensatory response in which neurons up-regulate glycolytic ATP production to compensate for declining mitochondrial respiration and OXPHOS energy production. In addition, decreased PDH function could also contribute to the correlated decrease in mitochondrial respiration and increase in glycolysis. PDH is the key rate-limiting enzyme in mitochondria to convert pyruvate, the end product of glycolysis, into acetyl-CoA, which subsequently condenses with oxaloacetate to initiate the TCA cycle for energy production. Compromised PDH function will lead to the accumulation of pyruvate and thus should stimulate anaerobic metabolism to lactic acid and cause an increase in extracellular acidification, as indicated by the increase in ECAR in neurons derived from 3xTg-AD neurons. Meanwhile, compromised PDH function in 3xTg-AD neurons leads to a deficit in acetyl-CoA and consequently decreased OXPHOS activity as indicated by the decrease in OCR in 3xTg-AD neurons. These findings suggest a potential antecedent role of mitochondrial bioenergetic deficits in AD pathogenesis and are consistent with previous positron emission tomography (PET) metabolic analyses in persons with increased risk of AD, mild cognitive impairment (MCI), or incipient to late AD, in which decreased glucose uptake and utilization was demonstrated as among the earliest symptoms of AD occurring far before the onset of AD (24–27). These findings are also consistent with microarray analyses of aging, incipient AD, and AD human samples and rodent models demonstrating that genes involved in mitochondrial bioenergetics are among those altered early in AD or 14674 兩 www.pnas.org兾cgi兾doi兾10.1073兾pnas.0903563106

MCI patients (22, 28). Consistent with mitochondrial dysfunction, decreased mitochondrial bioenergetics has been demonstrated to cause amyloid production and nerve cell atrophy (29, 30). In addition to compromised mitochondrial bioenergetics in 3xTg-AD mice, elevated oxidative stress was among the earliest indicators of dysfunction in the female 3xTg-AD mouse model. This finding is in agreement with previous studies from both the AD animal models and the human subjects that showed elevated oxidative stress plays an important role in AD pathogenesis (31, 32). The data also suggest that multiple aspects of mitochondrial dysfunction are closely related to the pathogenesis of AD. Oxidative damage to mitochondrial membranes and proteins is well documented to impair mitochondrial OXPHOS efficiency and result in increased electron leak as observed by increased hydrogen peroxide levels and higher oxidative stress (9). A direct link between A␤-induced toxicity and mitochondrial dysfunction in AD has been suggested by the interaction between mitochondrial A ␤ and a mitochondrial protein, ABAD (HSD17B10, 17␤-hydroxysteroid dehydrogenase) (20, 21). As in previous reports, our analyses show that the increase in mitochondrial A␤ correlated with the increase in ABAD level in the 3xTg-AD mouse brain. These observations confirmed the neurotoxic role of mitochondrial deposits of A␤. Nevertheless, the rise in ABAD and mitochondrial accumulation of A␤ was subsequent to the first symptom of mitochondrial dysfunction, which was the decline in mitochondrial bioenergetic activity. A decline in mitochondrial respiration and enzymes required for bioenergetics in vivo occurred in 3xTg-AD mice as early as 3 months of age. Supporting the hypothesis of mitochondrial dysfunction as an antecedent event to AD pathology, are the in vitro findings derived from embryonic neurons of 3xTg-AD mouse hippocampus. Embryonic hippocampal neurons exhibited significant deficits in mitochondrial respiratory function. Although we cannot rule out the possibility that the transgenes themselves could lead to mitochondrial dysfunction in the 3xTg-AD neurons, it is likely that decreased mitochondrial bioenergetics coupled with the increase in oxidative stress contributes to the overproduction of A␤, which when binding to ABAD forms an autocatalytic propagation of ROS further inducing mitochondrial dysfunction (33). In summary, mitochondrial dysfunction and deficits in bioenergetics occur early in pathogenesis and precede the development of observable plaque formation in female mouse model of AD. Further, age of reproductive senescence markedly exacerbated mitochondrial and bioenergetic dysfunction which is coincident with marked increases in AD pathology. If mitochondrial dysfunction is a causal link to Alzheimer’s, the susceptibility of mitochondria to environmental and genetic risks factors should be a critical factor in the development of late onset sporadic AD. This postulate is supported by the relationship between brain hypometabolism and increased risk of AD in offspring with maternal family history of AD (14, 27). As the mitochondrial genome is maternally inherited, this provides strong evidence of the potential causal role of mitochondrial dysfunction in AD pathogenesis. From a therapeutic perspective, the findings of this and other studies indicate a therYao et al.

Materials and Methods Transgenic Mice. Colonies of 3xTg-AD and nonTg mouse strain (C57BL/6/129S; The Jackson Laboratory) (16, 17) were bred and maintained at the University of Southern California (Los Angeles, CA). Details are provided in the SI Materials and Methods. Brain Tissue Preparation and Mitochondrial Isolation. Intact female mice of both 3xTg-AD and nonTg groups at different age groups (3, 6, 9, and 12 months) were killed. Brain mitochondria were isolated following a previously established protocol (36). Detailed methods are provided in the SI Materials and Methods. Immunohistochemistry. For immunohistochemistry study, animals were killed. Brains were perfused with prechilled PBS buffer and immersion fixed in 4% paraformaldehyde. Fixed brains were sent to Neuroscience Associates (NSA, Knoxville, TN) for coronal sectioning at 35 ␮m, and then processed for immunohistochemistry using a standard protocol. Detailed methods are provided in the SI Materials and Methods. Western Blot Analysis. Protein concentrations were determined using the BCA protein assay kit (Pierce). Western blot analysis was similar to a published procedure with minor modifications (36, 37). Detailed methods are provided in the SI Materials and Methods.

previously described (38). Detailed methods are provided in the SI Materials and Methods. Hydrogen Peroxide Production. The rate of hydrogen peroxide production by fresh isolated mitochondria was determined by the AmplexRed Hydrogen Peroxide/ Peroxidase Assay kit (Invitrogen) following the manufacturer’s instructions. Lipid Peroxidation. Lipid peroxides in brain mitochondria and hippocampal lysates were measured using the leucomethylene blue assay (5). Detailed methods are provided in the SI Materials and Methods. Respiratory Measurement. Mitochondrial oxygen consumption was measured polarographically using a Clark-type electrode following a previously published procedure (36, 37). Detailed methods are provided in the SI Materials and Methods. Free Radical Leak. Free radical leak was determined as previously described (36, 39). Detailed methods are provided in the SI Materials and Methods. Seahorse XF-24 Metabolic Flux Analysis. Primary hippocampal neurons from day 14 (E14) embryos of both 3xTg-AD and nonTg mice were cultured, and metabolic flux analysis of neurons derived from both 3xTg-AD and nonTg mice was performed according to the manufacturer’s instruction. Detailed methods are provided in the SI Materials and Methods. Statistics. Statistically significant differences between groups were determined by an ANOVA followed by a Newman-Keuls posthoc analysis.

Enzyme Activity Assay. PDH activity was measured by monitoring the conversion of NAD⫹ to NADH by following the change in absorption at 340 nm as

ACKNOWLEDGMENTS. This work was supported by Grants from the National Institute on Aging 1 PO1 AG026572 Progesterone in Brain Aging and Alzheimer’s Disease (to R.D.B.) and National Institutes of Mental Health 1RO1 MH67159 – 01 (to R.D.B.). R.W.I. was supported by National Institute on Aging training Grant T32-AG000093–24/25 (C.E. Finch, PI). Critical analysis provided by Dr. Enrique Cadenas is gratefully acknowledged.

1. Beal MF (2007) Mitochondria and neurodegeneration. Novartis Found Symp 287:183–192; discussion 192–196. 2. Duchen MR (2004) Mitochondria in health and disease: Perspectives on a new mitochondrial biology. Mol Aspects Med 25:365– 451. 3. Wallace DC (2008) Mitochondria as chi. Genetics 179:727–735. 4. Brinton RD (2008) The healthy cell bias of estrogen action: Mitochondrial bioenergetics and neurological implications. Trends Neurosci 31:529 –537. 5. Blass JP, Sheu RK, Gibson GE (2000) Inherent abnormalities in energy metabolism in Alzheimer disease. Interaction with cerebrovascular compromise. Ann N Y Acad Sci 903:204 –221. 6. Ishii K, et al. (1997) Reduction of cerebellar glucose metabolism in advanced Alzheimer’s disease. J Nucl Med 38:925–928. 7. Hoyer S (1991) Abnormalities of glucose metabolism in Alzheimer’s disease. Ann N Y Acad Sci 640:53–58. 8. Atamna H, Frey WH, 2nd (2007) Mechanisms of mitochondrial dysfunction and energy deficiency in Alzheimer’s disease. Mitochondrion 7:297–310. 9. Reddy PH, Beal MF (2008) Amyloid beta, mitochondrial dysfunction and synaptic damage: Implications for cognitive decline in aging and Alzheimer’s disease. Trends Mol Med 14:45–53. 10. King MP, Attardi G (1989) Human cells lacking mtDNA: Repopulation with exogenous mitochondria by complementation. Science 246:500 –503. 11. Khan SM, et al. (2000) Alzheimer’s disease cybrids replicate beta-amyloid abnormalities through cell death pathways. Ann Neurol 48:148 –155. 12. Swerdlow RH, et al. (1997) Cybrids in Alzheimer’s disease: A cellular model of the disease? Neurology 49:918 –925. 13. Cardoso SM, Santana I, Swerdlow RH, Oliveira CR (2004) Mitochondria dysfunction of Alzheimer’s disease cybrids enhances Abeta toxicity. J Neurochem 89:1417–1426. 14. Mosconi L, et al. (2009) Declining brain glucose metabolism in normal individuals with a maternal history of Alzheimer disease. Neurology 72:513–520. 15. Apostolova LG, et al. (2008) Subregional hippocampal atrophy predicts Alzheimer’s dementia in the cognitively normal. Neurobiol Aging Sep 24. [Epub ahead of print]. 16. Oddo S, Caccamo A, Kitazawa M, Tseng BP, LaFerla FM (2003) Amyloid deposition precedes tangle formation in a triple transgenic model of Alzheimer’s disease. Neurobiol Aging 24:1063–1070. 17. Oddo S, et al. (2003) Triple-transgenic model of Alzheimer’s disease with plaques and tangles: Intracellular Abeta and synaptic dysfunction. Neuron 39:409 – 421. 18. Carroll JC, et al. (2007) Progesterone and estrogen regulate Alzheimer-like neuropathology in female 3xTg-AD mice. J Neurosci 27:13357–13365. 19. Guppy M, Abas L, Arthur PG, Whisson ME (1995) The Pasteur effect in human platelets: Implications for storage and metabolic control. B J Haematol 91:752–757. 20. Yan SD, Stern DM (2005) Mitochondrial dysfunction and Alzheimer’s disease: Role of amyloid-beta peptide alcohol dehydrogenase (ABAD). Int J Exp Pathol 86:161–171. 21. Lustbader JW, et al. (2004) ABAD directly links Abeta to mitochondrial toxicity in Alzheimer’s disease. Science 304:448 – 452.

22. Blalock EM, et al. (2004) Incipient Alzheimer’s disease: Microarray correlation analyses reveal major transcriptional and tumor suppressor responses. Proc Natl Acad Sci USA 101:2173–2178. 23. Reiman EM, et al. (2004) Functional brain abnormalities in young adults at genetic risk for late-onset Alzheimer’s dementia. Proc Natl Acad Sci USA 101:284 –289. 24. Drzezga A, et al. (2003) Cerebral metabolic changes accompanying conversion of mild cognitive impairment into Alzheimer’s disease: A PET follow-up study. Eur J Nucl Med Mol Imaging 30:1104 –1113. 25. Jagust WJ, et al. (1991) Diminished glucose transport in Alzheimer’s disease: Dynamic PET studies. J Cereb Blood Flow Metab 11:323–330. 26. Cutler NR (1986) Cerebral metabolism as measured with positron emission tomography (PET) and [18F] 2-deoxy-D-glucose: Healthy aging, Alzheimer’s disease and Down syndrome. Prog Neuropsychopharmacol Biol Psychiatry 10:309 –321. 27. Liang WS, et al. (2008) Alzheimer’s disease is associated with reduced expression of energy metabolism genes in posterior cingulate neurons. Proc Natl Acad Sci USA 105:4441– 4446. 28. Rowe WB, et al. (2007) Hippocampal expression analyses reveal selective association of immediate-early, neuroenergetic, and myelinogenic pathways with cognitive impairment in aged rats. J Neurosci 27:3098 –3110. 29. Meier-Ruge WA, Bertoni-Freddari C (1997) Pathogenesis of decreased glucose turnover and oxidative phosphorylation in ischemic and trauma-induced dementia of the Alzheimer type. Ann N Y Acad Sci 826:229 –241. 30. Velliquette RA, O’Connor T, Vassar R (2005) Energy inhibition elevates beta-secretase levels and activity and is potentially amyloidogenic in APP transgenic mice: Possible early events in Alzheimer’s disease pathogenesis. J Neurosci 25:10874 –10883. 31. Sanz A, Pamplona R, Barja G (2006) Is the mitochondrial free radical theory of aging intact? Antioxid Redox Signal 8:582–599. 32. Shi Q, Gibson GE (2007) Oxidative stress and transcriptional regulation in Alzheimer disease. Alzheimer Dis Assoc Disord 21:276 –291. 33. Frederikse PH, Garland D, Zigler JS, Jr, Piatigorsky J (1996) Oxidative stress increases production of beta-amyloid precursor protein and beta-amyloid (Abeta) in mammalian lenses, and Abeta has toxic effects on lens epithelial cells. J Biol Chem 271:10169 –10174. 34. Simpkins JW, Singh M (2008) More than a decade of estrogen neuroprotection. Alzheimers Dement 4(1 Suppl 1):S131–S136. 35. Singh M, Sumien N, Kyser C, Simpkins JW (2008) Estrogens and progesterone as neuroprotectants: What animal models teach us. Front Biosci 13:1083–1089. 36. Irwin RW, et al. (2008) Progesterone and estrogen regulate oxidative metabolism in brain mitochondria. Endocrinology 149:3167–3175. 37. Nilsen J, Irwin RW, Gallaher TK, Brinton RD (2007) Estradiol in vivo regulation of brain mitochondrial proteome. J Neurosci 27:14069 –14077. 38. Gohil K, Jones DA (1983) A sensitive spectrophotometric assay for pyruvate dehydrogenase and oxoglutarate dehydrogenase complexes. Biosci Rep 3:1–9. 39. Sanz A, et al. (2005) Dietary restriction at old age lowers mitochondrial oxygen radical production and leak at complex I and oxidative DNA damage in rat brain. J Bionenerg Biomembr 37:83–90.

Yao et al.

PNAS 兩 August 25, 2009 兩 vol. 106 兩 no. 34 兩 14675

NEUROSCIENCE

apeutic strategy to prevent AD by sustaining mitochondrial metabolic function. Therapeutics, such as estrogen (4, 34, 35), that sustain and enhance mitochondrial functions by up-regulating key regulatory enzymes involved in brain metabolism, efficiency of mitochondrial bioenergetics whereas suppressing oxidative stress could prevent late onset AD.