Alzheimer’s disease is associated with reduced
expression of energy metabolism genes
in posterior cingulate neurons
Winnie S. Liang*†, Eric M. Reiman*†‡§, Jon Valla†¶, Travis Dunckley*†, Thomas G. Beach†?, Andrew Grover†?,
Tracey L. Niedzielko†¶, Lonnie E. Schneider†¶, Diego Mastroeni†?, Richard Caselli†**, Walter Kukull††, John C. Morris‡‡,
Christine M. Hulette§§, Donald Schmechel§§, Joseph Rogers†?, and Dietrich A. Stephan*†¶¶
*Neurogenomics Division, Translational Genomics Research Institute, 445 North Fifth Street, Phoenix, AZ 85004;†Arizona Alzheimer’s Consortium, 901 East
Willetta Street, Phoenix, AZ 85006;‡Banner Alzheimer’s Institute, 901 East Willetta Street Phoenix, AZ 85006;§Department of Psychiatry and Evelyn F.
McKnight Brain Institute, University of Arizona, 1501 North Campbell Avenue, Tucson, AZ 85724;¶Barrow Neurological Institute, 350 West Thomas
Road, Phoenix, AZ 85013;?Sun Health Research Institute, 10515 West Santa Fe Drive, Sun City, AZ 85351; **Department of Neurology, Mayo Clinic,
13400 East Shea Boulevard, Scottsdale, AZ 85259;††National Alzheimer’s Coordinating Center, 4311 11th Avenue NE, No. 300, Seattle, WA 98105;
‡‡Washington University Alzheimer’s Disease Research Center, Washington University School of Medicine, 4488 Forest Park Avenue, Suite 101,
St. Louis, MO 63108; and§§Bryan Alzheimer’s Disease Research Center, Duke University Medical Center, 2200 West Main Street, Suite A200,
Durham, NC 27705
Edited by Marcus E. Raichle, Washington University School of Medicine, St. Louis, MO, and approved January 15, 2008 (received for review
September 28, 2007)
Alzheimer’s disease (AD) is associated with regional reductions in
surements of the cerebral metabolic rate for glucose, which may
begin long before the onset of histopathological or clinical fea-
tures, especially in carriers of a common AD susceptibility gene.
Molecular evaluation of cells from metabolically affected brain
regions could provide new information about the pathogenesis of
AD and new targets at which to aim disease-slowing and preven-
tion therapies. Data from a genome-wide transcriptomic study
were used to compare the expression of 80 metabolically relevant
nuclear genes from laser-capture microdissected non-tangle-bear-
ing neurons from autopsy brains of AD cases and normal controls
in posterior cingulate cortex, which is metabolically affected in the
earliest stages; other brain regions metabolically affected in PET
studies of AD or normal aging; and visual cortex, which is relatively
spared. Compared with controls, AD cases had significantly lower
expression of 70% of the nuclear genes encoding subunits of the
mitochondrial electron transport chain in posterior cingulate cor-
tex, 65% of those in the middle temporal gyrus, 61% of those in
hippocampal CA1, 23% of those in entorhinal cortex, 16% of those
in visual cortex, and 5% of those in the superior frontal gyrus.
Western blots confirmed underexpression of those complex I–V
subunits assessed at the protein level. Cerebral metabolic rate for
glucose abnormalities in FDG PET studies of AD may be associated
with reduced neuronal expression of nuclear genes encoding
subunits of the mitochondrial electron transport chain.
gene expression ? Affymetrix microarrays ? laser capture micro-dissection
tomography (PET) measurements of the cerebral metabolic rate
for glucose (CMRgl). These CMRgl reductions have been
reported in the posterior cingulate, parietal, and temporal
affected patients (1–5). Other studies have reported CMRgl
reductions in anatomically well characterized hippocampal and
entorhinal cortical regions of interest (6–10). The posterior
cingulate cortex (PCC) and the neighboring precuneus are
metabolically affected in the earliest clinical and preclinical
stages of AD (4, 11), and the primary visual cortex is relatively
spared (4, 11). In an ongoing series of studies, we have detected
CMRgl reductions in cognitively normal carriers of the apoli-
poprotein E (APOE) ?4 allele (11–15), a common late-onset AD
susceptibility gene (16–18). CMRgl reductions in AD-affected
areas were correlated with APOE ?4 gene dose (i.e., three levels
lzheimer’s disease (AD) is associated with characteristic
and progressive reductions in regional positron emission
of genetic risk for AD) and were progressive in late-middle-aged
APOE ?4 heterozygotes (13), more than four decades before the
anticipated median onset of dementia, years before the expected
onset of the major histopathological features of AD (neurofi-
brillary tangles and amyloid plaques), and, indeed, in anticipat-
ing the initial regional appearance of fibrillar amyloid deposi-
tion. (20, 21).
AD-related CMRgl reductions could reflect reductions in the
density or activity of terminal neuronal fields or peri-synaptic
glial cells (22, 23), a metabolic dysfunction in neurons or glial
cells not related to neuronal activity (24, 25), or a combination
of these factors. These changes do not appear to be solely
attributable to the combined effects of atrophy and partial-
volume averaging (26). In a postmortem histochemistry study,
we previously found that AD cases had lower cytochrome c
oxidase activity than controls in the PCC, and that this reduction
was significantly greater than that in primary motor cortex (27),
another region that is relatively spared. Based on this observa-
tion, we proposed that AD might be related to an impairment in
Molecular evaluation of the cells from metabolically affected
brain regions could provide new information about the patho-
genesis of AD and new targets at which to aim disease-slowing
and prevention therapies. Data from a genome-wide transcrip-
tomic study were used to compare, in each of our sampled brain
regions, the expression of metabolically relevant nuclear genes
from laser-capture microdissected non-tangle-bearing neurons
of expired cases with clinically characterized and histopatholog-
ically verified AD and expired controls who did meet clinical
criteria for dementia and histopathological criteria for AD.
In particular, the data were used to compare cases and
controls in the expression of 80 nuclear genes encoding mito-
chondrial electron transport chain (ETC) subunits along with
translocases of the inner and outer mitochondrial membranes
Author contributions: D.A.S. designed research; W.S.L., J.V., T.D., T.G.B., T.L.N., and L.E.S.
performed research; A.G., D.M., R.C., W.K., J.C.M., C.M.H., D.S., and J.R. contributed new
The authors declare no conflict of interest.
This article is a PNAS Direct Submission.
Data deposition: The data reported in this paper have been deposited in the Gene
Expression Omnibus (GEO) database, www.ncbi.nlm.nih.gov/geo (accession no. GSE5281).
© 2008 by The National Academy of Sciences of the USA
March 18, 2008 ?
vol. 105 ?
no. 11 ?
(TIMMs and TOMMs, respectively), in six brain regions. These
nuclear genes included 39 complex I genes coding for NADH
dehydrogenase, all 4 nuclear-encoded complex II genes coding
for succinate dehydrogenase, 9 complex III genes coding for
ubiquinol-cytochrome c reductase, 13 complex IV genes coding
for cytochrome c oxidase, and 15 complex V genes coding for
ATP synthase, as well as 11 TIMMs and 6 TOMMs, which
regulate the transport of nuclear-encoded electron transport
subunits into the mitochondria. These ETC complexes and
translocases are illustrated in Fig. 1. The six brain regions
included the PCC, which is associated with unusually early
CMRgl reductions in AD (4, 28); the middle temporal gyrus
(MTG); the hippocampal field CA1 (HIP) and entorhinal cortex
(EC), which are also metabolically affected in AD (1, 20); the
superior frontal gyrus (SFG), which is associated with prefer-
ential CMRgl reductions in normal aging (29–33); and the
primary visual cortex (VC), which is relatively spared from
CMRgl reductions in both aging and AD (4, 11).
We initially tested the hypothesis that AD would be associated
with reduced neuronal expression of metabolically relevant
nuclear genes in the PCC and that these AD-related reductions
would be significantly greater than those in VC. (Findings from
our survey of 55,000 neuronal transcripts have been published in
ref. 54.) AD-related reductions in the posterior cingulate neu-
ronal expression of several of the implicated complexes I–IV and
ATP synthase (complex V) subunits were subsequently validated
at the protein level, suggesting that the reductions in the
neuronal expression of metabolically relevant nuclear genes
might be associated with the CMRgl reductions found in flu-
orodeoxyglucose positron emission tomography (FDG PET)
studies of AD.
Fold change values and P values for the 80 ETC and translocase
genes in each of the sampled regions are posted at www.tgen.org/
neurogenomics/pcc, and additional expression results have been
reported in ref. 54. The proportion of significantly underex-
pressed ETC and translocase genes in each of the sampled brain
regions (P ? 0.01 after correction for multiple comparisons) is
shown in Table 1. In comparison with controls, AD cases had
significantly lower expression of 70% of the nuclear genes
encoding mitochondrial ETC subunits in PCC pyramidal neu-
rons, 65% of those in MTG neurons, 61% in HIP neurons, 23%
in EC layer II stellate neurons, 16% in VC neurons, and 5% in
SFG neurons. By comparison, AD cases had significantly higher
expression of only 4% of the nuclear genes encoding mitochon-
drial ETC subunits in PCC pyramidal neurons, 13% of those in
MTG neurons, 8% of HIP neurons, 1% of EC neurons, 4% of
VC neurons, and 0% of SFG neurons (data for these genes are
listed on the supplementary data site). Additional factors that
displayed statistically significant underexpression in the PCC,
and which may influence mitochondrial energy metabolism,
include the TIMMs and TOMMs (Table 1; individual P values
and fold changes are located on the supplementary data site),
which are required for the transportation of proteins, including
ETC components, from the cytoplasm to the inner mitochon-
drial membrane and different compartments in the mitochon-
dria (34–37). Overall, 35% of these translocases demonstrated
pression in the PCC are shown. These elements include the five complexes of the ETC and TIMMs and TOMMs. OM, outer mitochondrial membrane; IMS,
intermembrane space; IM, inner mitochondrial membrane.
Altered expression of mitochondrial energy metabolism elements. Energy metabolism-relevant elements showing statistically significant underex-
Table 1. The proportion of underexpressed metabolism-related genes in each of the sample brain regions
Complex IComplex IIComplex III Complex IVComplex VTIMMs TOMMs
The numerator indicates the number of subunits showing statistically significant (P ? 0.01 with multiple testing corrections) underexpression and the
denominator indicates the total number of nuclear-encoded subunits for the complex/translocase. This ratio is expressed as a percentage in parentheses. PCC,
posterior cingulate cortex; MTG, middle temporal gyrus; HIP, hippocampus; EC, entorhinal cortex; VC, visual cortex; SFG, superior frontal gyrus).
www.pnas.org?cgi?doi?10.1073?pnas.0709259105Liang et al.
underexpression in the PCC neurons, 47% in the MTG neurons,
41% in the HIP neurons, 35% in the EC neurons, 6% in the VC
neurons, and 24% in the SFG neurons. By comparison, only 6%
of the translocases showed overexpression in the PCC and HIP;
12% in the VC; and 0% in the MTG, EC, and SFG.
In a post hoc analysis, we compared the proportion of all genes
associated with underexpression of 61% of statistically signifi-
cant (P ? 0.01, corrected for multiple comparisons) genes in the
PCC, 43% of those in the MTG, 56% of those in the HIP, 69%
of those in the EC, 81% of those in the SFG, and 58% of those
in the VC. These findings suggest that although metabolism
genes may not be disproportionately affected in AD, the regional
pattern of underexpressed metabolic genes (PCC ? SFG and
VC) more closely reflects the pattern of metabolic reductions
observed in PET studies.
To validate our findings of underexpression of energy meta-
bolic factors, we performed Western blots on the posterior
cingulate in controls and AD cases. Concurrent with expression
findings, statistically significant decreases were found in the
protein levels of each enzyme complex subunit assessed using
two-tailed t tests, P ? 0.01): complex I (mitochondrially encoded
subunit ND6), 41.9 ? 7.0% (mean ? 95% CI) of control, P ?
III, 45.3 ? 7.8% of control, P ? 0.008; complex IV (mitochon-
drially encoded subunit COXII), 59.1 ? 9.3% of control, P ?
0.002; and complex V, 72.4 ? 11.1% of controls, P ? 0.000008
(Fig. 2; human heart was used as positive control)
This study provides transcriptomic and protein evidence of neuro-
nal metabolic impairments in AD brains, complementing previ-
ously established PET evidence of metabolic impairments in per-
sons afflicted by or at risk for this disorder. In our analysis of the
nuclear genes influencing mitochondrial energy metabolism (i.e.,
80 nuclear genes encoding ETC subunits and 17 nuclear genes
encoding mitochondrial translocases responsible for the entry of
ETC subunits into the mitochondria), the largest proportion of
underexpressed genes was in the PCC, a region which PET studies
find to be metabolically affected in the earliest stages of AD. The
proportion of underexpressed genes was significantly greater than
those in the VC, which is relatively spared in PET studies of AD.
The MTG, EC, and HIP, which are also affected in PET studies of
AD, had proportions of underexpressed genes in between those in
the PCC and VC.
Because our analysis was confined to laser-capture microdis-
sected neurons, our findings suggest that the CMRgI reductions
observed in PET studies of AD are at least partly related to
molecular processes in neurons themselves, and are thus not
solely attributable to a reduction in the activity of perisynaptic
the CMRgI reductions could be at least partly related to
molecular processes in neuronal cell bodies (i.e., changes in
nuclear gene expression), even though alterations in PET CM-
RgI measurements have been suggested to be more strongly
influenced by the activity of terminal neuronal fields (22).
Furthermore, this differential pattern of ETC and metabolic
involvement across the cortex highlights the differential vulner-
ability of various regions to the pathophysiology of AD. Our
current transcriptomic findings and previous PET studies raise
the possibility that the molecular processes involved in neuronal
energy metabolism may be involved in the earliest pathogenesis
of AD, possibly preceding the onset of neuritic plaques.
Earlier gene expression and functional activity studies provide
confirmatory evidence of this regional pattern of ETC changes.
Underexpression was found for mitochondrially-encoded and
nuclear-encoded ETC subunits in the middle temporal cortex,
but not in motor cortex, of AD patients (38, 39). In a laminar
analysis, the PCC also showed the most prominent and signifi-
cant functional declines across each cortical layer in complex IV
activity, whereas the motor cortex was relatively spared (27).
Middle temporal, superior temporal, and inferior parietal cor-
tices showed significant layer III activity declines, in agreement
with our gene expression data, with the midfrontal cortex
showing the smallest decline (40). Further, as the layer III cell
bodies from which the expression data were taken project apical
dendrites (containing the highest metabolic demand) into the
superficial layers of the cortex, these layers (I and II) showed
globally, albeit not always significantly, reduced ETC activity
that could result from decreased subunit expression in the layer
III soma (40).
ETC subunits, which have been validated here with Western
blots, corresponds to the regional pattern of CMRgI reductions
observed in PET studies of patients with AD (28, 41–45) such
that the PCC and precuneus are preferentially affected in AD
(11, 16, 17, 19). Because PET studies have also identified the
PCC as showing the earliest metabolic changes in cognitively
normal carriers of the APOE ?4 allele (19), these changes may
precede downstream development of AD pathologies including
amyloid plaque and neurofibrillary tangle formation. Metabolic
alterations in the PCC have also been implicated in maintaining
the brain’s ‘‘default’’ state when it is not engaged in the perfor-
mance of specific tasks (46) or episodic memory tasks (20), as
well as in the predisposition to initial hypometabolic (11–13) and
subsequent fibrillar amyloid changes (20, 47, 48) associated with
AD. The current study provides new information about the
neuronal processes involved in these normal and pathological
This study capitalized on the genome-wide evaluation of gene
transcripts from laser-capture microdissected neurons from ex-
tremely high-quality brain tissue [mean postmortem interval
(PMI) ? 2.5 h] in clinically and neuropathologically well-
characterized AD cases and age-matched controls, and on the
mixture labeling a subunit of each mitochondrial enzyme complex from PCC
whole-brain extract. H, human heart for positive control; AD, Alzheimer’s
right. From highest to lowest are complex V alpha subunit, complex III core 2
subunit, complex II 30-kDa subunit, complex IV subunit II, and complex I
20-kDA subunit. Results are representative of duplicate blots. (B) Western
blot-band optical density, expressed as percentage of normal-aged control.
for each subunit tested. n ? 10 AD; n ? 5 NC.
Western blot validation. (A) Western blots using a five-antibody
Liang et al.
March 18, 2008 ?
vol. 105 ?
no. 11 ?
analysis of neurons from brain regions preferentially affected or
spared in PET studies of AD. Although this study provides clues
about the molecular processes related to the CMRgI reductions
observed in persons afflicted by and at genetic risk for AD, it has
several limitations. First, given the descriptive nature of our
findings, it remains to be clarified whether the changes in the
nuclear expression of mitochondrial metabolism genes cause
reductions in the density or activity of terminal neuronal fields
or instead are a consequence of the reduced metabolic demands
associated with terminal neuronal changes in AD. Second, it also
remains to be clarified whether or not AD is also associated with
underexpression of mitochondrially encoded genes from the
same neurons (a technically challenging question to address in
laser-capture microdissected cells) or in the nuclear or mito-
chondrial genes of perisynaptic glial cells. Third, although we
controlled for mean age at death and gender, there is a small
possibility that other group differences in the agonal state
preceding death contributed to some of the AD-related reduc-
tions in neuronal gene expression observed in this study. Finally,
findings from our study suggest that a high proportion of genes
are underexpressed in AD, whether or not they are known to
directly influence neuronal metabolism. However, the regional
pattern of underexpressed genes regulating mitochondrial me-
tabolism appear to correspond to the pattern of CMRgI changes
Noting the pattern of metabolic changes in our PET studies of
AD, Roses et al. have suggested that insulin sensitizers may be
helpful in the treatment and prevention of AD (49, 50). Findings
from our neurotranscriptomic study could provide clues about
the molecular mechanisms that may be involved in the earliest
pathogenesis of this disorder and potential targets at which to
aim new disease-slowing and prevention therapies.
This study provides transcriptomic and protein evidence that the
neuronal nuclear genes influencing mitochondrial energy me-
tabolism are underexpressed in AD, particularly in brain regions
like the PCC, which are found to be preferentially affected in
PET studies of AD patients and cognitively normal persons at
genetic risk for this disorder. In doing so, this study provides
information about the molecular processes involved in the
pathogenesis of AD and potential therapeutic targets at which to
aim disease-slowing and prevention therapies.
Materials and Methods
Tissue Collection. Brain samples were collected at three Alzheimer’s Disease
Centers (Washington University, Duke University, and Sun Health Research
Institute) from clinically and neuropathologically classified late-onset AD-
afflicted individuals (15 males and 18 females) with a mean age of 79.9 ? 6.9
years. Tissue collection of healthy elderly controls was published in a previous
report (51); Consortium to Establish a Registry for Alzheimer’s Disease
I to II. All individuals were Caucasian and were matched as closely as possible
logically or metabolically relevant to AD. These include the EC (BA 28 and 34),
SFG (BA 10 and 11), HIP, VC (BA 17), MTG (BA 21 and 37), and PCC (BA 23 and
31). The EC and HIP are preferentially affected by intracellular neurofibrillary
tangles, the MTG and PCC are preferentially affected by metabolism and
extracellular ?-amyloid plaques, the SFG is preferentially affected by aging,
and the VC is relatively spared from both aging and AD pathologies. After
dissection, samples were frozen, sectioned (10 ?m), and mounted on glass
As previously described (53), brain sections were stained with a combina-
tion of thioflavin-S (Sigma) and 1% neutral red (Fisher Scientific), pyramidal
the region of interest, and tangles were identified by the bright green
fluorescence of thioflavin-S staining. In the EC, the large stellate neurons
lacking thioflavin-S staining were collected from layer II, and pyramidal cells
lacking thioflavin-S staining were collected from CA1 of the HIP. The CA1
region was selected for study because this area is the most, and earliest,
affected region in terms of tangle formation, and it has already been expres-
sion profiled in neurologically healthy elderly individuals. In all other regions,
(for all collected neurons, cell bodies were extracted). For each individual,
?500 histopathologically normal pyramidal neurons were collected from the
six brain regions by using laser-capture microdissection (Veritas automated
laser-capture microdissection system; Arcturus). Cells were collected onto
Arcturus CapSure Macro LCM caps and extracted according to the manufac-
turer’s instructions. Total RNA was isolated from the cell lysate by using the
Arcturus PicoPure RNA isolation kit with DNase I treatment using the Qiagen
RNase-free DNase set.
Expression Profiling. Expression profiling was performed as previously de-
scribed (51). Isolated total RNA was double-round amplified, cleaned, and
a T7 promoter and the Ambion MEGAscript T7 high-yield transcription kit
quantitated on a spectrophotometer and run on a 1% TAE gel to check for an
evenly distributed range of transcript sizes. Twenty micrograms of cRNA was
fragmented to ?35–200 bp by alkaline treatment (200 mM Tris-acetate, pH
8.2; 500 mM KOAc; 150 mM MgOAc) and run on a 1% TAE gel to verify
fragmentation. Separate hybridization cocktails were made by using 15 ?g of
fragmented cRNA from each sample according to Affymetrix’s instructions.
Microarray Analysis. Two hundred microliters of each mixture was separately
hybridized to an Affymetrix Human Genome U133 Plus 2.0 array for 16 h at
45°C in the Hybridization Oven 640. The Affymetrix human genome arrays
measure the expression of over 47,000 transcripts and variants, including
38,500 characterized human genes. Hybridization cocktails for nine EC sam-
profiling methodology used in this project (53) were reanalyzed on the U133
Plus 2.0 array to be evaluated in this study; a 10th EC sample was also
separately processed for this sample group. Arrays are washed on the Af-
fymetrix upgraded GeneChip Fluidics Station 450 by using a primary strepta-
vidin phycoerythrin (SAPE) stain, subsequent biotinylated antibody stain, and
3000 7G with AutoLoader. Scanned images obtained by the Affymetrix Ge-
neChip Operating Software (GCOS) v1.2 were used to extract raw signal
intensity values per probe set on the array and calculate detection calls
(absent, marginal, or present). Assignment of detection calls is based on
probe-pair intensities for which one probe is a perfect match of the reference
sequence and the other is a mismatch probe for which the 13th base (of the
25-oligonucleotide reference sequence) is changed. All raw chip data were
scaled to 150 in GCOS to normalize signal intensities for interarray compari-
sons. Reports generated by GCOS were reviewed for quality control; we
a scaling factor ?10. Twenty arrays that failed to pass these standards were
not included in further analyses.
Pyramidal Cell Quality Control. To ensure neuronal cell purity in the samples,
we evaluated expression of GFAP, an astrocyte cell marker. Fourteen samples
were removed from statistical analyses.
were generated in a previous study (51). Microarray data files of the normal
samples are available on the Gene Expression Omnibus site at www.ncbi.nlm.
nih.gov/geo/query/acc.cgi?acc ? GSE5281 (project accession no. GSE5281), and
regional analyses are posted at www.tgen.org/neurogenomics/data/private3.
We compared AD cases and controls in the neuronal expression of 80 nuclear
genes encoding subunits of the mitochondrial ETC in each of our sampled brain
regions. The nuclear genes included 39 complex I genes coding for NADH dehy-
NDUFS1–S8, and NDUFV1–V3), all 4 complex II genes coding for succinate dehy-
drogenase (SDHA, SDHB, SDHC, and SDHD), 9 complex III genes coding for
ubiquinol-cytochrome c reductase (UCRC, UQCR, UQCRB, UQCRC1, UQCRC2,
c oxidase (COX4I1, COX5A, COX5B, COX6A1, COX6A2, COX6B1, COX6C,
COX7A2L, COX7B, COX7B2, COX7C, COX8A, and COX8C), and 15 complex V
ATP5G1–G3, ATP5H, ATP5I, ATP5J, ATP5J2, ATP5L, and ATP5O). TIMM and
TOMM expression was also evaluated (TIMM8A, TIMM8B, TIMM9, TIMM10,
TIMM13, TIMM17A, TIMM17B, TIMM22, TIMM23, TIMM44, TIMM50, TOMM7,
www.pnas.org?cgi?doi?10.1073?pnas.0709259105Liang et al.
encoded subunits were not assessed in this study.
Direct comparisons between neurologically healthy and AD-afflicted brains
listed ETC subunits. For each analysis, genes that did not demonstrate at least
?10% present calls across all transcripts profiled for each region-specific com-
parison were removed by using Genespring GX 7.3 Expression Analysis software
(with a multiple testing correction using the Benjamini and Hochberg false
present-call criterion to locate genes that were statistically significant in differ-
entiating expression between healthy brains and AD brains. (After the present-
in the EC, 32,265 genes in the MTG, 31,496 genes in the HIP, 32,482 genes in the
VC, and 32,118 genes in the SFG.) For each analysis comparing AD expression
levels with control levels, genes that had a corrected P value ? 0.01 were
collected, and those genes whose average AD signal and average control signal
were both below a threshold of 150 were removed. Fold-change values were
scaled expression signal for the same gene from the normal samples across all
Using this approach, we evaluated ETC and translocase genes that were
differentially expressed in the AD cases versus controls for each of the six brain
and overexpression were determined by calculating the number of subunits (at
total number of nuclear-encoded subunits (at least ?10% present calls with
corrected P ? 0.01) on the Affymetrix human genome array. Evaluation of all
human genes demonstrating underexpression in each region focused on only
those genes with statistical significance (P ? 0.01 with multiple testing correc-
tions, after present-call filters) for each regional AD versus controls comparison;
underexpression percentages were calculated based on a ratio of underex-
pressed statistically significant genes over all genes.
Data Posting. Minimum information about a microarray experiment (MIAME)-
compliant microarray data files for control samples are located on the Gene
Expression Omnibus (GEO) site at www.ncbi.nlm.nih.gov/geo/query/
acc.cgi?acc?GSE5281 (project accession no. GSE5281). Fold changes and cor-
rected and uncorrected P values for all ETC and translocase genes on the Af-
fymetrix Human Genome Array for each of the six regional comparisons are
available online at www.tgen.org/neurogenomics/pcc. Posted lists show respec-
will be published in a separate report.
5 profiled healthy control cases were collected for Western blot validation.
from slides by using PBS, and 35 ?g of protein per lane was loaded into
10–20% 1.5 mm Tris-glycine 15-well minigels (Invitrogen) with Novex Tris-
90 min at constant 150 V with Novex Tris-glycine SDS running buffer. Proteins
were transferred to Immobilon P (Millipore) PVDF membrane, fully sub-
merged in CAPS transfer buffer (10 mM 3-[cyclohexylamino]-1-propane sul-
Membranes were blocked overnight in 5% Carnation dry milk in PBS at 4°C
and then probed with the total OXPHOS detection kit (MS601; 1:1,000 in 1%
dry milk; Mitosciences) for 2 h at room temperature with gentle rocking. This
drially encoded), the 30-kDa iron–sulfur subunit of complex II, the core 2
F1? subunit of ATP synthase. Goat anti-mouse alkaline phosphatase (AP)-
conjugated secondary antibody (1:10,000 in 1% dry milk; Santa Cruz Biotech-
nology) was used (2 h at room temperature with rocking), followed by
application of an AP substrate detection kit (Bio-Rad), according to the
manufacturer’s instructions. Membranes were dried overnight, then simulta-
in duplicate. The image was imported into Optimas image analysis software
(Media Cybernetics), and average OD measures of each band were taken by
using a sampling window of constant size. Groups were compared by using
Student’s two-tailed t tests, uncorrected for multiple comparisons, and 95%
confidence intervals for each complex were calculated in Excel.
ACKNOWLEDGMENTS. We thank Roger Higdon and Daniel McKeel for their
support through Alzheimer’s Disease Research Centers, Lucia Sue (Sun Health
Research Institute) for assistance with clinical data, Nick Lehmans (Transla-
tional Genomics Research Institute) for setting up the supplementary data
site, the National Institute on Aging’s Alzheimer’s Disease Centers (ADC)
program, and the National Alzheimer’s Coordinating Center for help in
obtaining samples for analysis. This project was funded by the National
P30 AG19610 (to E.M.R. for Arizona ADC), P50 AG05681 (to J.C.M.), P01
AG03991 (to J.C.M.), AG05128 (for the Duke University Alzheimer’s Disease
Research Center), National Alzheimer’s Coordinating Center Grant
U01AG016976, the state of Arizona (E.M.R.), and the Barrow Neurological
Foundation Women’s Board (J.V.).
1. Alexander GE, Chen K, Pietrini P, Rapoport SI, Reiman EM (2002) Longitudinal PET
Alzheimer’s disease treatment studies. Am J Psychiatry 159:738–745.
2. Fox NC, Cousens S, Scahill R, Harvey RJ, Rossor MN (2000) Using serial registered brain
magnetic resonance imaging to measure disease progression in Alzheimer disease:
Power calculations and estimates of sample size to detect treatment effects. Arch
3. Jack CR, Jr, et al. (2004) Comparison of different MRI brain atrophy rate measures with
clinical disease progression in AD. Neurology 62:591–600.
early Alzheimer’s disease. Ann Neurol 42:85–94.
5. Thal LJ, et al. (2006) The role of biomarkers in clinical trials for Alzheimer disease.
Alzheimer Dis Assoc Disord 20:6–15.
2-[18F]fluoro-2-deoxy-D-glucose/positron-emission tomography (FDG/PET). Proc Natl
Acad Sci USA 98:10966–10971.
7. Mosconi L, et al. (2007) Hippocampal hypometabolism predicts cognitive decline from
normal aging. Neurobiol Aging, 10.1016/j.neurobiol.aging.2006.12.008.
in MCI, AD. Neurobiol Aging 22:529–539.
9. Desgranges B, et al. (1998) The neural substrates of memory systems impairment in
Alzheimer’s disease. A PET study of resting brain glucose utilization. Brain 121:611–
rate decreases in hippocampal formation and proisocortex than in neocortex in
Alzheimer’s disease. Neuropsychobiology 37:10–19.
11. Reiman EM, et al. (1996) Preclinical evidence of Alzheimer’s disease in persons ho-
mozygous for the epsilon 4 allele for apolipoprotein E. N Engl J Med 334:752–758.
12. Reiman EM, et al. (2001) Declining brain activity in cognitively normal apolipoprotein
E epsilon 4 heterozygotes: A foundation for using positron emission tomography to
efficiently test treatments to prevent Alzheimer’s disease. Proc Natl Acad Sci USA
13. 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.
14. Small GW, et al. (2000) Cerebral metabolic and cognitive decline in persons at genetic
risk for Alzheimer’s disease. Proc Natl Acad Sci USA 97:6037–6042.
in relatives at risk for familial Alzheimer disease. J Am Med Assoc 273:942–947.
16. Corder EH, et al. (1993) Gene dose of apolipoprotein E type 4 allele and the risk of
Alzheimer’s disease in late onset families. Science 261:921–923.
17. Saunders AM, et al. (1993) Association of apolipoprotein E allele epsilon 4 with
late-onset familial and sporadic Alzheimer’s disease. Neurology 43:1467–1472.
18. Tang MX, et al. (1998) The APOE-epsilon4 allele and the risk of Alzheimer disease
among African Americans, whites, and Hispanics. J Am Med Assoc 279:751–755.
19. Reiman EM, et al. (2005) Correlations between apolipoprotein E epsilon4 gene dose
20. Buckner RL, et al. (2005) Molecular, structural, and functional characterization of
memory. J Neurosci 25:7709–7717.
marker of Alzheimer disease. Neurology 67:446–452.
22. Schwartz WJ, et al. (1979) Metabolic mapping of functional activity in the hypo-
thalamo-neurohypophysial system of the rat. Science 205:723–725.
23. Magistretti PJ, Pellerin L (1996) Cellular bases of brain energy metabolism and their
relevance to functional brain imaging: Evidence for a prominent role of astrocytes.
Cereb Cortex 6:50–61.
and phosphorylation in Alzheimer’s disease determined by dynamic FDG-PET. J Nucl
25. Mark RJ, Pang Z, Geddes JW, Uchida K, Mattson MP (1997) Amyloid beta-peptide
impairs glucose transport in hippocampal and cortical neurons: Involvement of mem-
brane lipid peroxidation. J Neurosci 17:1046–1054.
26. Ibanez V, et al. (1998) Regional glucose metabolic abnormalities are not the result of
atrophy in Alzheimer’s disease. Neurology 50:1585–1593.
27. Valla J, Berndt JD, Gonzalez-Lima F (2001) Energy hypometabolism in posterior cin-
gulate cortex of Alzheimer’s patients: Superficial laminar cytochrome oxidase associ-
ated with disease duration. J Neurosci 21:4923–4930.
Liang et al.
March 18, 2008 ?
vol. 105 ?
no. 11 ?
28. Minoshima S, Foster NL, Kuhl DE (1994) Posterior cingulate cortex in Alzheimer’s Download full-text
disease. Lancet 344:895.
29. Angelie E, et al. (2001) Regional differences and metabolic changes in normal aging of the
human brain: proton MR spectroscopic imaging study. AJNR Am J Neuroradiol 22:119–127.
30. Convit A, et al. (2001) Volumetric analysis of the pre-frontal regions: Findings in aging
and schizophrenia. Psychiatry Res 107:61–73.
31. Ivancevic V, et al. (2000) Regional cerebral glucose metabolism in healthy volunteers
determined by fluordeoxyglucose positron emission tomography: appearance and
variance in the transaxial, coronal, and sagittal planes. Clin Nucl Med 25:596–602.
32. Loessner A, et al. (1995) Regional cerebral function determined by FDG-PET in healthy
volunteers: Normal patterns and changes with age. J Nucl Med 36:1141–1149.
34. Bauer MF, Hofmann S, Neupert W, Brunner M (2000) Protein translocation into
mitochondria: the role of TIM complexes. Trends Cell Biol 10:25–31.
Nat Rev Mol Cell Biol 2:339–349.
37. Berthold J, et al. (1995) The MIM complex mediates preprotein translocation across the
mitochondrial inner membrane and couples it to the mt-Hsp70/ATP driving system. Cell
expression in Alzheimer disease. Brain Res Mol Brain Res 24:336–340.
39. Chandrasekaran K, Hatanpaa K, Rapoport SI, Brady DR (1997) Decreased expression of
nuclear and mitochondrial DNA-encoded genes of oxidative phosphorylation in asso-
ciation neocortex in Alzheimer disease. Brain Res Mol Brain Res 44:99–104.
istry in human Alzheimer’s disease and animal models. J Histotechnology 30:1–30.
41. McGeer EG, et al. (1990) 18Fluorodeoxyglucose positron emission tomography studies
in presumed Alzheimer cases, including 13 serial scans. Can J Neurol Sci 17:1–11.
42. Mielke R, Herholz K, Grond M, Kessler J, Heiss WD (1994) Clinical deterioration in
probable Alzheimer’s disease correlates with progressive metabolic impairment of
association areas. Dementia 5:36–41.
43. Minoshima S, Frey KA, Koeppe RA, Foster NL, Kuhl DE (1995) A diagnostic approach in
Alzheimer’s disease using three-dimensional stereotactic surface projections of fluo-
rine-18-FDG PET. J Nucl Med 36:1238–1248.
44. Pietrini P, et al. (1993) Pattern of cerebral metabolic interactions in a subject with
isolated amnesia at risk for Alzheimer’s disease: A longitudinal evaluation. Dementia
45. Smith GS, et al. (1992) Topography of cross-sectional and longitudinal glucose meta-
bolic deficits in Alzheimer’s disease. Pathophysiologic implications. Arch Neurol
46. Raichle ME, et al. (2001) A default mode of brain function. Proc Natl Acad Sci USA
47. Braak H, Braak E (1991) Neuropathological stageing of Alzheimer-related changes.
Acta Neuropathol 82:239–259.
48. Mirra SS, et al. (1991) The Consortium to Establish a Registry for Alzheimer’s Disease
(CERAD). Part II standardization of the neuropathologic assessment of Alzheimer’s
disease. Neurology 41:479–486.
49. Risner ME, et al. (2006) Efficacy of rosiglitazone in a genetically defined population
with mild-to-moderate Alzheimer’s disease. Pharmacogenomics J 6:246–254.
50. Roses AD, et al. (2007) Complex disease-associated pharmacogenetics: Drug efficacy,
drug safety, and confirmation of a pathogenetic hypothesis (Alzheimer’s disease).
Pharmacogenomics J 7:10–28.
51. Liang WS, et al. (2007) Gene expression profiles in anatomically and functionally
distinct regions of the normal aged human brain. Physiol Genomics 28:311–
52. McKhann G, et al. (1984) Clinical diagnosis of Alzheimer’s disease: Report of the
NINCDS-ADRDA work group under the auspices of Department of Health and Human
Services Task Force on Alzheimer’s disease. Neurology 34:939–944.
53. Dunckley T, et al. (2006) Gene expression correlates of neurofibrillary tangles in
Alzheimer’s disease. Neurobiol Aging 27:1359–1371.
affected by Alzheimer’s disease: A reference data set. Physiol Genomics, 10.1152-
www.pnas.org?cgi?doi?10.1073?pnas.0709259105 Liang et al.