Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2011 Aug 1.
Published in final edited form as: Mol Psychiatry. 2010 May 18;16(8):836–847. doi: 10.1038/mp.2010.57

Molecular signatures in post-mortem brain tissue of younger individuals at high risk for Alzheimer’s disease as based on APOE genotype

C Conejero-Goldberg 1, TM Hyde 2, S Chen 1, U Dreses-Werringloer 1, MM Herman 2, JE Kleinman 2, P Davies 1,3, TE Goldberg 1,4
PMCID: PMC2953572  NIHMSID: NIHMS212621  PMID: 20479757

Abstract

Alzheimer’s disease (AD) is a neurodegenerative condition characterized histopathologically by neuritic plaques and neurofibrillary tangles. The objective of this transcriptional profiling study was to identify both neurosusceptibility and intrinsic neuroprotective factors at the molecular level, not confounded by the downstream consequences of pathology. We thus studied post-mortem cortical tissue in 28 cases that were non-APOE4 carriers (called the APOE3 group) and 13 cases that were APOE4 carriers. As APOE genotype is the major genetic risk factor for late-onset AD, the former group was at low risk for development of the disease and the latter group was at high risk for the disease. Mean age at death was 42 years and none of the brains had histopathology diagnostic of AD at the time of death. We first derived interregional difference scores in expression between cortical tissue from a region relatively invulnerable to AD (primary somatosensory cortex, BA 1/2/3) and an area known to be susceptible to AD pathology (middle temporal gyrus, BA 21). We then contrasted the magnitude of these interregional differences in between-group comparisons of the APOE3 (low risk) and APOE4 (high risk) genotype groups. We identified 70 transcripts that differed significantly between the groups. These included EGFR, CNTFR, CASP6, GRIA2, CTNNB1, FKBPL, LGALS1 and PSMC5. Using real-time quantitative PCR, we validated these findings. In addition, we found regional differences in the expression of APOE itself. We also identified multiple Kyoto pathways that were disrupted in the APOE4 group, including those involved in mitochondrial function, calcium regulation and cell-cycle reentry. To determine the functional significance of our transcriptional findings, we used bioinformatics pathway analyses to demonstrate that the molecules listed above comprised a network of connections with each other, APOE, and APP and MAPT. Overall, our results indicated that the abnormalities that we observed in single transcripts and in signaling pathways were not the consequences of diagnostic plaque and tangle pathology, but preceded it and thus may be a causative link in the long molecular prodrome that results in clinical AD.

Keywords: AD, APOE, microarray, gene expression, human brain

Introduction

Alzheimer’s disease (AD) is a devastating neurodegenerative disorder that afflicts ~5 million Americans, most of whom are elderly. One underappreciated approach to developing therapeutic strategies comes from findings that have observed clear differences in the severity of neuropathology between different cortical brain regions in AD. Primary somatosensory cortex (BA 1/2/3) remains relatively uncompromised until late in the disease process, whereas lateral temporal cortex, including the middle temporal gyrus (BA 21), becomes involved at relatively early stages.1-4 The mechanism by which this occurs has remained elusive. The primary aim of this study was to elucidate differences in gene expression profiling between brain regions susceptible to AD pathology and regions that are relatively spared. In principle, this would allow greater understanding of both intrinsic protective and susceptibility mechanisms in individuals at high risk for the disease.

As we were interested in identifying changes in expression that might be associated with early neuroprotective and/or pathogenic responses and not the consequences of AD pathology, we studied postmortem brain tissue of individuals who did not have diagnosable AD neuropathology (that is, plaques and tangles), but nevertheless were at high risk to develop AD on the basis of their APOE genotype. APOE is the largest and best replicated genetic risk factor for late-onset AD.5 The disadvantageous E4 allele is associated with an odds ratio of 3.5–4 for AD in heterozygotes and 12–15 in homozygotes. Thus, we could classify cases into two groups: a low-risk group comprised of non-APOE4 carriers and a high-risk group comprised of APOE4 carriers.

Several previous studies attempted to identify disease-related factors in brain tissue from AD cases; many were subjected to a variety of conceptual and technical criticisms. We were able to address and overcome these problems that have made interpretation of earlier results problematic:6,7 mRNA variability (high RNA integrity number (RIN), cases and controls ≥7); pH (mean = 6.6); differential medication history between cases and controls (cases were not on psychotropic or anti-Alzheimer’s medications); agonal state (death was not protracted); validation (using real-time quantitative PCR, RT-qPCR); statistical issues, including corrections for multiple comparisons and use of permutation analyses; downstream consequences of late-onset end-stage AD (cases did not have AD-related plaques and tangles, but, by intention, were at markedly elevated risk for AD). Several studies used relatively small sample sizes; to our knowledge, the sample size in this study is the largest in the preclinical AD literature.

Materials and methods

Human brain samples and platform

Human brain tissue from normal control subjects (N = 41) without any history of neuropsychiatric illness, neurological disease or drug abuse was obtained from the Clinical Brain Disorders Branch of the NIMH. The gray matter from BA 21 and BA 1/2/3 were obtained for each subject using a high-speed hand-held dental drill. Demographic characteristics of this cohort were as follows: age (years) = 41.9±10.3 (range 24–59); sex = 34M/7F; pH = 6.6±0.29; post-mortem interval (PMI) (hours) = 31.9±16.4; RNA-RIN: BA21 = 8.0±0.5 and BA1/2/3 = 8.3±0.5. We note that RNA quality is more closely related to RIN and pH than PMI.8,9 In our sample, RIN and pH were equivalent and high in both APOE groups.

Of the 41 brain samples used for the microarray study, genotyping determined that 24 cases were E3 homozygotes and four cases were E3/E2 heterozygotes. These 28 cases constituted a non-APOE4 group, called hereafter the APOE3 group. Thirteen cases were APOE4 carriers (eleven E4/E3 heterozygotes and two E4 homozygotes).

Neither age, PMI, pH, nor critically RIN differed between the APOE3 and APOE4 groups by t-test (all P-values > 0.23). An imbalance in sex was observed (six females in the E3 group, one in the E4 group), leading us to block by sex in our analyses of variance (Supplementary Materials and Methods).

Tissue samples from frozen blocks were pulverized and total RNA was extracted using standard procedures as previously described.10 Briefly, total RNA (DNAse treated) was isolated using an RNeasy kit with a QIAshredder column (Qiagen, Valencia, CA, USA) and measured for quality using Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, USA). No RNA samples with RIN values < 7.0 were included in this study. Microarrays were used according to the manufacturer’s guidelines (Illumina, San Diego, CA, USA). Total RNA (260 ng) was converted to cDNA by reverse transcription using ArrayScript reverse transcriptase and T7-oligo (dT)24 primers, followed by second-strand synthesis to generate double-stranded cDNA. After purification, the cDNA was converted to biotin-labeled cRNA (Totalprep RNA Labeling Kit, Ambion, Austin, TX, USA), hybridized to the microarray platform: HumanWG-6_V2 Expression BeadChip (Illumina), and stained with streptavidin-Cy3 for visualization. Two technical replicates with independent cDNA and cRNA synthesis steps were used per brain sample. A total of 164 microarrays were run (41 samples × 2 regions × 2 technical replicates).

Pathology

Routine neuropathology did not reveal evidence of AD (amyloid deposits, neuritic plaques or neurofibrillary tangles) in these samples. In all cases, a neuropathological examination was carried out on tissue samples taken from the frontal, temporal and occipital lobes, parietal lobe convexity and cerebellar vermis using hematoxylin–eosin and Bielschowsky’s silver impregnation adapted for paraffin-embedded sections. In cases suggestive of AD or other dementias, either by clinical history, gross appearance of the brain or advanced age, we also examined a sample of the caudal hippocampus with the same stains. The criteria applied for the neuropathological diagnosis of AD were age adjusted and based on those of Khachaturian;11 see Supplementary Materials and Methods for details.

Genotyping

Total DNA was extracted from the cerebellum tissue of the same patients as previously described and genotyped.10,12 As noted, genotyping determined that 24 cases were APOE3 homozygotes and 4 cases were E3/E2 heterozygotes (the APOE3 group) and 13 cases were APOE4 carriers (11 cases were E4/E3 heterozygotes and 2 cases E4 homozygotes), the APOE4 group.

Data pre-processing

After the probe arrays were scanned, the resulting images were first pre-processed using the BeadStudio software (Illumina), which calculates the mean fluorescence signal across all 30 replicates of each gene/transcript (AVG_Signal), along with a detection score that represents the probability that the mean signal for each gene/transcript on the chip is greater than background (that is, detection P-value). Genes/transcripts were defined as being significantly expressed above background (as detected by the array) when each gene’s detection P-value was ≤0.001. The expression data were then normalized within quantiles across samples of the distribution of gene expression values. The quantile normalization method was used to make the distribution of probe intensities equivalent for all samples.

BeadStudio calculates background as the average signal intensity estimated from the negative-control bead types (~700) and removes outliers using the median absolute deviation method (Illumina BeadStudio). However, previous studies13 andour ownpilot studies indicated that background subtraction had a negative impact on data quality (for example, it lowered correlation coefficients between technical replicates), so we therefore exported data that were normalized, but that did not undergo background subtraction. As a result of this processing, 26 357 of 48 701 possible transcripts met criteria and were used in all subsequent analyses in BRB Array Tools 3.7 developed by NIH (http://www.linus.nci.nih.gov/BRB-ArrayTools.html).

Statistical analysis

Normalized data were processed using a ‘double subtraction’ method. First, interregional differences were computed for each transcript for each case. Thus, expression values from BA 1/2/3 were subtracted from expression values from BA 21. These difference scores were then imported to BRB Array Tools and processed within the univariate module. Technical replicates of the same sample were averaged. The APOE3 and APOE4 groups were compared and contrasted on these difference scores (hence ‘double subtraction’). Like-lihood ratio test statistics (F-test) were used to investigate the significance of the difference score between the classes, that is, an F-test was computed separately for each transcript using the normalized intensities. We report P-values for the class variable only.

Expression differences were considered statistically significant if P < 0.001. Our procedures for limiting the proportion of false discoveries involved a multivariate permutation test in which the false discovery rate was set at 0.20 as detailed in Supplementary Materials and Methods.

In our primary analysis, we examined differences in untransformed signal intensity between regions (BA 21 v. BA 1/2/3) within subjects, followed by a class comparison (that is, E3 v. E4) blocked by sex in BRB Array Tools. By doing so, we sought to determine whether the magnitude of interregional differences within subjects differed between APOE3 and APOE4 genotypes. If such class comparison were present, this would suggest the presence of neurosuscetibility and/or neuroprotective responses. Thus, we conducted a genotype by region interaction analysis.

Cluster analysis of transcript profiles

A cluster analysis of gene expression patterns was conducted in SAS version 9.1.3 (Cary, NC, USA; Proc Cluster) as an unbiased and unsupervised method for determining patterns of expression across genes. As we were interested in the profiles of APOE3 temporal cortex and primary somatosensory cortex signal intensities and APOE4 temporal and primary somatosensory cortex signal intensities (four variables), not absolute intensity values, all intensity values were converted to z-scores based on each transcript’s grand mean. Cubic Clustering Criterion ( > 2) and the difference between estimated and actual variance explained by the cluster solution (differences in favor of actual) were used to determine the number of clusters and their composition (that is, which transcripts were included in each cluster).

MDS: clustering of cases

In order to determine separation of APOE3 and APOE4 carriers at the level of individual cases, multidimensional scaling (MDS) analysis was conducted. Of the transcripts that significantly differed, a principal components analysis was conducted and three orthogonal components were identified.14 Cases were then assigned to loci in three-dimensional space based on individualized component scores. MDS was performed using Euclidean distance as a distance metric on uncentered data. A likelihood test was conducted in order to determine the probability that two or more clusters were present, as compared with the null hypothesis.

Hierarchical clustering of expression changes

Hierarchical clustering of transcript difference scores was performed using Euclidean distance as a distance metric and complete linkage. The cluster analysis of transcripts (and cases) produces a heat map image in which the rows in the image plot represent the genes, and the columns in the image plot represent the brain samples. Thus, this reflects case profiles in terms of transcript difference scores.

Biological and signaling pathways analyses (Kyoto Encyclopedia of Genes and Genomes)

LS actual P-values were based on random sampling of P-values of N probes among all probes in the array in comparison to the P-values of the N probes in the specified path, as were Kolmogorov-Smirnov (KS) based values. Interregional difference scores of signal intensity comprised the raw values that were subjected to pathways analyses; APOE genotype comprised the class comparison from which the P-value derived for individual transcripts within the prespecified path. In all, 100 000 permutations were performed for each pathway.

RT-qPCR

Validation of selected transcripts was conducted using RT-qPCR. For selected transcripts (12 genes) showing differential expression between the APOE4 and APOE3 groups, cDNA synthesis was generated for each sample with the Ambion reverse transcription Kit and oligo dT primer. For each sample, amplified product differences were measured with two replicates with locked nucleic acid chemistrybased detection. For housekeeping genes, primers and probes used, see Supplementary Materials and Methods. The RT-qPCR reactions were carried out in an ABI Prism 7900HT thermal cycler (Applied Biosystems, Foster City, CA, USA), determining the ΔCt and significance of the RT-qPCR-reported APOE3-APOE4 group differences. Statistical analysis on BA 21-BA 1/2/3 difference scores was performed using a Student’s t-test in Microsoft Excel. Thus, we recapitulated the analytic procedures in our microarray study, but here we used gene product determined by RT-qPCR.

Functional significance of molecules for APP and MAPT processing: IPA

We ascertained the functional significance of susceptibility and protective molecules to amyloid and tau pathology using Ingenuity Pathway Analysis (IPA; Ingenuity Systems, http://www.ingenuity.com).15 We first entered our molecules of interest: the nine transcripts that we validated in RT-qPCR, as well as APOE. Fold changes for each transcript were also included in order to refine the specificity of the analysis. The transcripts served as seeds for generating networks that maximize their specific connectivity. Additional molecules from the IPA database were used to merge smaller networks into larger ones, as selected by an IPA algorithm. The IPA ‘build’ algorithm was used to grow connections between the network and MAPT and amyloid β precursor (APP), proteins associated with AD pathology. A Network Value score was then generated, indicating the likelihood that the network connectivity was found by chance.

Results

Transcripts

We identified 70 transcripts corresponding to genes from RefSeq and UniGene databases that differed significantly (at P < 0.001) between the APOE3 and APOE4 carriers using ‘double subtraction’ analyses. The global P-value for the gene set analysis was significant (P = 0.04). In all, 51 of these transcripts survived false discovery rate analyses. The 70 transcripts, their exact P-values and difference score are listed in Table 1. We provide descriptive information about several of these transcripts (also validated in RT-qPCR, see section on RT-qPCR) and their relevance to AD in Supplementary Results.16-39

Table 1.

Transcripts that discriminate among classes

GenBank
accesion		 Gene symbol/
UniGene
cluster		 Description Geometric mean
intensities
class E4		 Geometric mean
intensities
class E3		 Fold
change		 Parametric
P-value
NM_003112 SP4 Sp4 transcription factor (SP4), mRNA 1.16 1.02 1.13 0.00001
NM_144597 C15orf40 Chromosome 15 open reading frame 40, mRNA 1.14 1.00 1.14 0.00002
AI798812 Hs.523978 cDNA clone IMAGE:2348622 3, mRNA 1.07 0.98 1.09 0.00004
AV681468 Hs.164146 cDNA clone GKAAAF07 5, mRNA 1.07 0.97 1.10 0.00006
NM_001008395 LOC389541 Similar to CG14977-PA, mRNA 0.91 1.04 0.88 0.00006
NM_002281 KRT81 Keratin 81 (KRT81), mRNA 0.93 1.01 0.92 0.00006
NM_207173 NPSR1 Neuropeptide S receptor 1, variant 2, mRNA 0.92 1.00 0.92 0.00009
AA017152 Hs.103355 cDNA clone IMAGE:361525 3, mRNA 0.93 1.01 0.92 0.00010
CN315248 Hs.543452 cDNA 5, mRNA 1.05 0.95 1.10 0.00012
BM703456 Hs.528873 cDNA clone UI-E-CL1-afe-e-04-0-UI 5, mRNA 1.14 1.01 1.12 0.00013
NM_019896 POLE4 Polymerase (DNA-directed), epsilon 4, mRNA 0.92 1.07 0.86 0.00013
DB230039 Hs.580513 cDNA clone TRACH3023310 5, mRNA 0.94 1.02 0.92 0.00015
NM_001226 CASP6 Caspase 6, variant α and β 1.11 0.98 1.13 0.00018
NM_007358 MTF2 Metal response element binding transcription
factor 2		 1.17 0.99 1.18 0.00018
NM_178550 C1orf110 Chromosome 1 open reading frame 110, mRNA 1.10 1.01 1.10 0.00019
NM_139174 LOC161931 Nuclear RNA-binding protein-like, mRNA 0.69 0.89 0.78 0.00019
NM_001546 ID4 Inhibitor of DNA binding 4, mRNA 1.29 1.06 1.21 0.00020
NM_020727 ZNF295 Zinc-finger protein 295, mRNA 1.08 0.94 1.15 0.00020
NM_000337 SGCD Sarcoglycan, delta, variant 1, mRNA 1.20 1.06 1.13 0.00021
NM_147164 CNTFR Ciliary neurotrophic factor receptor, variant 1,
mRNA		 1.09 0.99 1.11 0.00021
NR_000010 SNORD4A Small nucleolar RNA, C/D box 4A 1.10 0.98 1.12 0.00022
XR_001273 LOC647357 PREDICT: similar to CG14980-PB, variant 1,
RNA		 1.11 1.00 1.11 0.00022
XM_379203 LOC348801 PREDICT: hypothetical protein, variant 1,
mRNA		 1.07 1.00 1.07 0.00022
XM_497712 LOC441907 PREDICT: similar to 60S ribosomal protein L6,
mRNA		 1.07 0.97 1.10 0.00023
NM_015364 LY96 Lymphocyte antigen 96, mRNA 0.84 0.97 0.87 0.00023
BQ003813 Hs.555565 cDNA clone IMAGE:5847799 3, mRNA 0.95 1.02 0.93 0.00024
XR_018067 LOC441228 PREDICT: similar to exportin-T, mRNA 0.89 0.99 0.90 0.00026
BX393727 Hs.440088 cDNA clone CS0DC001YP02 5-PRIME, mRNA 1.05 0.94 1.11 0.00033
XM_926149 LOC642384 PREDICT: hypothetical protein LOC642384,
mRNA		 1.15 1.03 1.12 0.00035
NM_018433 JMJD1A Jumonji domain containing 1A, mRNA 1.08 0.95 1.14 0.00036
NM_004326 BCL9 B-cell CLL/lymphoma 9, mRNA 1.05 0.96 1.09 0.00038
NM_139168 SFRS12 Splicing factor, arginine–serine-rich 12, variant
2, mRNA		 1.10 0.97 1.14 0.00038
NM_004890 SPAG7 Sperm-associated antigen 7, mRNA 0.93 1.03 0.90 0.00040
BF590392 Hs.542085 cDNA clone IMAGE:3258165 3, mRNA 0.95 1.03 0.91 0.00041
NM_016202 ZNF580 Zinc-finger protein 580, transcript variant 1,
mRNA		 1.10 1.01 1.08 0.00041
AI935042 Hs.575097 cDNA clone IMAGE:2328437 3, mRNA 1.03 0.97 1.07 0.00042
BX089938 Hs.130107 cDNA clone IMAGE:1504127, mRNA 1.05 0.96 1.10 0.00043
NM_080390 TCEAL2 Transcription elongation factor A (SII)-like 2,
mRNA		 1.06 0.99 1.07 0.00043
XM_496731 LOC441056 PREDICT: similar to double homeobox 4c,
mRNA		 0.94 1.02 0.93 0.00044
BE439917 Hs.539307 HTM1-532F HTM1, cDNA, mRNA 0.91 1.00 0.92 0.00045
NM_006190 ORC2L Origin recognition complex, subunit 2-like,
mRNA		 0.94 1.01 0.93 0.00045
NM_000312 PROC Protein C, mRNA 0.98 1.06 0.92 0.00046
NM_079423 MYL6 Myosin, light chain 6, transcript variant 2,
mRNA		 0.92 1.01 0.91 0.00047
BX093261 Hs.98049 cDNA clone IMAGE:753731, mRNA 0.96 1.03 0.93 0.00047
NM_004146 NDUFB7 NADH dehydrogenase 1 β subcomplex, 7,
mRNA		 0.95 1.03 0.92 0.00049
NM_030815 PDRG1 p53 and DNA damage regulated 1, mRNA 0.92 1.03 0.90 0.00049
NM_002805 PSMC5 Proteasome 26S subunit, ATPase, 5, mRNA 0.92 0.98 0.94 0.00049
BX476711 Hs.121070 cDNA clone DKFZp686M05188 5, mRNA 1.12 0.98 1.14 0.00051
NM_178867 SFXN4 Sideroflexin 4, transcript variant 2, mRNA 1.05 0.96 1.09 0.00052
NM_153822 PSMD4 Proteasome 26S subunit, 4, variant 2, mRNA 0.90 1.00 0.90 0.00056
XM_938229 PIAS2 PREDICT: protein inhibitor of activated STAT-
2, mRNA		 1.09 0.96 1.14 0.00058
NM_002305 LGALS1 Lectin, galactoside-binding, soluble, 1, mRNA 0.65 0.84 0.78 0.00060
NM_022171 TCTA T-cell leukemia translocation-altered gene,
mRNA		 0.94 1.04 0.91 0.00061
BX115738 Hs.551137 cDNA clone IMAGp998D07589, mRNA 1.05 0.97 1.08 0.00063
XM_940109 LOC650995 PREDICT: similar to Golgi autoantigen, golgin
subfamily a, 2		 0.96 1.05 0.91 0.00066
AW016203 Hs.566210 cDNA clone IMAGE:2712316 3, mRNA 0.95 1.02 0.93 0.00067
XM_943679 WBSCR19 PREDICT: Williams–Beuren syndrome
chromosome 19, variant 4		 1.11 0.98 1.14 0.00070
NM_145237 LOC94431 Similar to RNA polymerase I transcription
factor RRN3, mRNA		 1.06 0.98 1.08 0.00071
NM_001012753 ZNF763 Zinc-finger protein 763, mRNA 1.11 0.99 1.12 0.00073
NM_022110 FKBPL FK506 binding protein-like (FKBPL), mRNA 0.95 1.03 0.93 0.00074
NM_012262 HS2ST1 Heparan sulfate 2-O-sulfotransferase 1, mRNA 1.25 1.10 1.14 0.00075
NM_080678 UBE2F Ubiquitin-conjugating enzyme E2F (putative),
mRNA		 0.90 1.01 0.89 0.00075
NM_003789 TRADD TNFRSF1A-associated via death domain,
mRNA		 0.80 0.89 0.89 0.00077
NM_005530 IDH3A Isocitrate dehydrogenase 3 (NAD+) α, mRNA 0.77 0.87 0.88 0.00077
NM_001029862 ANKRD30B Ankyrin repeat domain 30B, mRNA 1.16 1.01 1.16 0.00078
AI651118 Hs.565928 cDNA clone IMAGE:2304126 3, mRNA 0.94 1.01 0.92 0.00084
NM_001011537 FYTTD1 Forty-two-three domain-containing 1,
variant 2, mRNA		 1.05 0.98 1.07 0.00085
NM_004930 CAPZB Capping protein (actin filament) Z-line, β,
mRNA		 0.92 1.04 0.89 0.00087
NR_003264 SDHALP1 Succinate dehydrogenase complex, subunit A 1.11 0.98 1.14 0.00088
BI561597 Hs.576019 cDNA clone IMAGE:5298552 5, mRNA 0.97 1.03 0.94 0.00092
Open in a new tab

Abbreviation: CLL, chronic lymphocytic leukemia.

Seventy transcripts are significant at the nominal 0.001 level of the univariate test. Sorted by P-value of the univariate test. Class 1: E4; class 2: E3.

We also directly compared APOE3 BA 21 transcripts with APOE4 BA 21 transcripts and APOE3 BA 1/2/3 with APOE4 BA1/2/3 transcripts (that is, ‘single subtraction’). Results are in Supplementary Material.

Cluster analysis of 70 gene transcripts identified by analysis of variance

To determine whether consistent profile expression patterns could be observed in this set of transcripts, we conducted a cluster analysis using the average distance method of the signal intensity values of the 70 genes identified previously in the univariate analyses. Four clusters were determined (Cubic Clustering Criterion = 7.56, estimated variance = 0.82, actual variance = 0.90, Figures 1a–d). APOE4 BA 21 signal intensities were upregulated in clusters 1 and 4 and downregulated in clusters 2 and 3, as might be anticipated if susceptibility related changes in this vulnerable region were occurring. Strikingly, APOE4 BA 1/2/3 values were upregulated in clusters 2 and 3 and downregulated in clusters 1 and 4 in a type of reciprocal relationship to BA 21 findings, possibly indicating the presence of constitutive, regionally specific, neuroprotective adaptations. Cluster 1, in which BA 21 transcripts genes were upregulated and BA 1/2/3 transcripts were downregulated, contained the largest number of transcripts and hence was the modal pattern. Representative transcripts from cluster 1 (CASP6) and cluster 2 (FKBPL) are shown in Figures 1e and f. Transcripts that comprised each cluster are listed in Supplementary Table 1.

Figure 1.

Figure 1

Open in a new tab

Cluster analysis of transcript patterns. (a–d) An unsupervised cluster analysis of transcript profiles yielded four patterns. Use of z-scores controlled for effects of absolute levels of signal. (a) Cluster 1 was modal and involved upregulation of the transcript in APOE4 BA 21 temporal cortex and downregulation in APOE4 BA 1/2/3 primary somatosensory cortex; expression levels in these two regions were not strikingly different in the APOE3 group. (b–c) The next most frequent pattern involved upregulation of expression in primary somatosensory cortex of the APOE4 group, with downregulation in the temporal cortex; differences between these regions in the E3 group were relatively smaller. (d) This pattern was least frequent. (e) CASP6, an example of a transcript included in cluster1. CASP6 is involved in programmed cell death. It is strongly upregulated in the temporal cortex of the APOE4 group, while modestly downregulated in the primary somatosensory cortex of the APOE4 group. This may represent a response in the APOE4 group that confers susceptibility to AD pathology. (f) FKBPL, an example of a transcript from cluster 2. FKBPL is an immunophilin that has neuroprotective properties. It was strongly upregulated in BA 1/2/3 (primary somatosensory cortex) of the APOE4 group and modestly downregulated in BA 21 (temporal cortex) of the APOE3 group, reflecting a possible neuroprotective response in BA 1/2/3. Units on the y axis represent expression in signal intensity units. Along the x axis are temporal and parietal variables, as a function of APOE group.

Genotype group separation

MDS

MDS analysis (equivalent to principal components analysis) of the first three orthogonal components derived from the significant transcripts accounted for 0.59 of the total variance. The existence of two clusters (APOE3 and APOE4) was supported at a strong trend level of significance (P = 0.06) and by visual inspection of the cloud (Supplementary Figure 1). Of two cases distant from the APOE3 cloud, one was an APOE4 homozygote.

Hierarchical clustering of expression changes

As shown in the heat map image (Figure 2), the hierarchical clustering of expression changes demonstrated that in the majority of cases in the APOE4 group (9 of 13) about 55% (37 of 70) of the significant transcripts identified by univariate statistics demonstrated interregional differences in expression, which were associated with strong upregulation in BA 21 lobe and downregulation in BA 1/2/3 (that is, ‘red’ cells). The APOE3 group demonstrated a less distinct pattern of upregulation and downregulation, suggesting that interregional differences were less robust in this group. This was consistent with the parametric analyses above.

Figure 2.

Figure 2

Open in a new tab

Heat map showing regional up- and downregulations on a case-by-case basis for the 70 transcripts identified as significant. A hierarchical clustering algorithm was applied to the expression data. The 70 transcripts identified as significant in the univariate analysis (all P-values < 0.001) are on the y axis and cases, grouped according to their similarity to their neighbors, are on the x axis. Log intensities of temporal/primary somatosensory cortex ratios are represented in colored cells. Red cells indicate relatively strong upregulation in BA 21 (temporal cortex) and downregulation in BA 1/2/3 (primary somatosensory cortex), whereas green cells indicate the converse. An APOE4 homozygote was the right-most case.

Pathway analyses

Biological and signaling pathways

Altogether, 22 Kyoto pathways were identified with LS P-values of P < 0.00001 (Table 2). These included several of the pathways that have been implicated in previous studies of AD, including wnt signaling, calcium signaling, cell cycle, insulin signaling, oxidative phosphorylation, neuroactive ligands and receptors, and proteasome function. Nominally significant transcripts (P < 0.05) within all pathways are listed in Supplementary Table 2 and include transcripts that we examined in RT-qPCR, namely, GRIA2, EGFR and CNNTB1. The pathways and their relevance to AD are discussed in greater detail in Supplementary Results, Transcripts and Signaling Pathways.40-50 Gene Ontology results are in Supplementary Results.

Table 2.

Kyoto Encyclopedia of Genes and Genomes pathways that differ significantly between the APOE3 and APOE4 groups

Pathway description Number of
transcripts
(P < 0.05) 		Number of
genes
in pathway		 LS permutation
P-value 		KS permutation
P-value
Oxidative phosphorylation (mitochondrial function) 30 124 0.00001 0.00001
Purine metabolism 18 167 0.00001 0.00001
Glycan structures—biosynthesis 1 7 113 0.00001 0.00001
Proteasome 30 33 0.00001 0.00009
Calcium signaling pathway 16 183 0.00001 0.00001
Cytokine–cytokine receptor interaction 15 210 0.00001 0.00001
Neuroactive ligand–receptor interaction 22 251 0.00001 0.00001
Cell cycle 10 123 0.00001 0.00001
Wnt signaling pathway 21 161 0.00001 0.00001
Axon guidance 10 153 0.00001 0.00001
Focal adhesion 21 235 0.00001 0.00001
Cell adhesion molecules 15 127 0.00001 0.00001
Adherens junction 14 101 0.00001 0.00001
Tight junction 13 140 0.00001 0.00001
Gap junction 15 114 0.00001 0.00001
Jak-STAT signaling pathway 10 135 0.00001 0.00001
Natural killer cell-mediated cytotoxicity 12 116 0.00001 0.00001
Leukocyte transendothelial migration 13 123 0.00001 0.00001
Regulation of actin cytoskeleton 25 234 0.00001 0.00001
Insulin signaling pathway 16 151 0.00001 0.00001
Gonadotropin-releasing hormone signaling pathway 11 113 0.00001 0.00001
Colorectal cancer 15 104 0.00001 0.00001
Open in a new tab

KS, Kolmogorov-Smirnov.

RT-qPCR

We validated our microarray findings in 9 transcripts that were chosen from the 70 transcripts that differed between the APOE groups and from the Kyoto signaling pathways: EGFR, CNTFR, CASP6, ZNF580, GRIA2, CTNNB1, FKBPL, LGALS1 and PSMC5. These transcripts were chosen on the basis of their statistical significance, role in key biological or signaling pathways, and relevance to AD pathogenesis. They are all listed in Supplementary Tables 3 and 4, along with P-value significance levels, fold changes and fold changes determined by our microarray. Critically, for each positive finding, we found that the pattern of RT-qPCR results was consistent with that of the microarray results in terms of regional up- or downregulation for the APOE3 and APOE4 groups. A comparison of fold changes between the microarray and RT-qPCR data is in Supplementary Figure 2; detailed graphs of the corresponding microarray expression values for positive transcripts are in Supplementary Figure 3.

APOE expression

Owing to the pivotal role of APOE genotype in our study, we conducted additional statistical analyses of expression using repeated-measures analysis of variance (conducted in SAS version 9.1.3; Proc GLM). Genotype was considered a class variable and region an intrasubject repeated factor. We determined that no main effect of genotype was present (this was thus consistent with the double subtraction results) and no genotype by region interaction was present. Critically, however, a main effect of region was present (such that expression of both APOE isoforms was greater in BA 21 than in BA 1/2/3 (F = 46.72, P = 0.002), see Supplementary Figure 4a. Collapsed across APOE groups, mean expression in the temporal lobe in signal intensity units was 6836±3348 s.e.m.; in the parietal lobe, it was 5040±3193 s.e.m. This observation may be fundamental in understanding why cells in this region (BA 1/2/3) were able to mount presumptively neuroprotective responses, but not in temporal lobe. We validated this finding in RT-qPCR. There was an interregional difference of expression of APOE (collapsed across APOE groups) that was significant by matched pair t-test (t = 2.47, P = 0.02) and consistent with the microarray data, such that expression was lower in the parietal lobe than in the temporal lobe (Supplementary Figure 4b).

Functional significance of findings: IPA

To examine the functional significance of our findings, we conducted a network analysis of the interrelationships among the nine molecules that we validated in RT-qPCR, and APP and MAPT (tau), the proteins associated with amyloid and tau aggregation in end-stage AD. We used IPA. Results shown in Figure 3 from IPA indicate that there are multiple direct and indirect paths from the nine molecules that we previously identified to both MAPT and APP. Eight of the nine molecules were considered Network Eligible: EGFR, CNTFR, CASP6, GRIA2, CTNNB1, FKBPL, LGALS1, PSMC5, as well as APOE. The overall network strength value was 25 (that is, P <1 × 10−24), indicating that the strength of the relationships among the molecules was unlikely to be a chance finding. We also identified two hub molecules (mean number of connections per molecules+2 s.d.): CTNNB1 and EGFR. They are discussed in the Supplementary Results.

Figure 3.

Figure 3

Open in a new tab

Gene network illustration demonstrating multiple interrelationships among eight key transcripts that were validated in RT-qPCR experiments and were found to be network eligible by IPA. Transcripts shown in red were upregulated in BA 21 (temporal cortex) in the APOE4 group and were considered susceptibility related; transcripts shown in green were upregulated in BA 1/2/3 (primary somatosensory cortex) and were considered neuroprotective in the APOE4 group. Edges indicate multiple direct and indirect relationships among the transcripts and their character. Other relevant genes added by IPA algorithm to connect smaller networks into larger networks are also shown. Most strikingly, connections between the key transcripts and APP, misprocessing of which results in amyloid aggregation, and MAPT, related to tau fibrillization, were also identified. Two of the transcripts were also identified as hub molecules: CTNNB1 and EGFR. These may have especially important roles in AD pathogenesis. A network legend is provided in the Supplementary Materials.

Discussion

Our results yielded four primary findings. First, the earliest impetus for pathogenic processes in APOE4 individuals may come from a variety of abnormalities in signaling cascades and biological processes that involve calcium regulation, mitochondrial function, cell-cycle regulation abnormalities, apoptosis and wnt signaling. Several of these have been at the periphery of discussions about the pathogenesis of AD; in our study, they were found to be central. Second, there may be active protective processes as well as pathological processes. Third, our data suggest that there is a long prodromal period before the onset of clinical AD, given that mean age at death of our sample was 42 years. Finally, we identified lower levels of APOE transcript in BA 1/2/3 compared with BA 21. Thus, even in individuals who carry an APOE4 allele, relatively low amounts of this isoform are present in BA 1/2/3. Speculatively, this may account for the region’s relative invulnerability to histopathology, either because expression levels are not high enough to promote pathogenesis or to interfere with neuroprotective adaptations. In contrast, conformation of the E4 variant including its structural and biophysical features,51,52 in conjunction with relatively high expression levels, may result in increased pathology in susceptible regions.

The MDS clustering also discriminated APOE3 and APOE4 cases, thus indicating that even in this early stage in a putative disease process, separation at the level of individual cases seemed to be present. The heat map data were consistent with this interpretation. Transcripts that are included as significant in these analyses were EGFR, CNTFR, CASP6, ZNF580, GRIA2, CTNNB1, FKBPL, LGALS1 and PSMC5. These transcripts were not only statistically highly significant, but were in key biological or signaling pathways that have relevance to AD pathogenesis. Analyses in IPA were consistent with this view. We found that the transcripts that we validated in RT-qPCR had multiple direct and indirect relationships (‘edges’) among themselves and APP and MAPT processing, strongly suggesting that they affect the disease process.

We identified 22 signaling pathways and biological processes that differed between the APOE3 and APOE4 groups. Several may be of special interest because they have previously been implicated in AD, including oxidative phosphorylation/mitochondrial function, calcium signaling, cell-cycle reentry, neuroactive ligand–receptor interactions, wnt signaling and proteasome functionality. Our study suggests that they are involved in early, presumptively pathogenic features of the disease process. Our results in mitochondrial function were particularly striking. Although mitochondria have become an area of increasing importance in the study of AD and other dementias,53-56 it has generally been assumed that amyloid plaques or toxic amyloid fragments interfere with mitochondrial function.57 Our results suggest that compromises in mitochondrial function may precede plaque formation, in keeping with experiments suggesting that mitochondrial dysfunction may precede amyloid aggregation.56 In our study, the transcripts found to be most different between the APOE3 and APOE4 groups in this pathway were consistently downregulated in temporal lobe in the APOE4 group and included NDUF7. This is consistent with previous microarray work on mitochondrial function in AD brain tissue, that found abnormalities in complexes I, II, III, IV and V.58 Nevertheless, it is likely that later in the disease process amyloid misprocessing amplifies mitochondrial dysfunction.59

Although we did not identify molecules directly associated with amyloid processing, we are not suggesting that the amyloid cascade is not relevant for AD. There is overwhelming evidence for its importance from highly penetrant genetic mutations, transgenic animals, neuropathology and molecular experiments.60 Rather, we think that the data in this study suggest that in APOE4-related AD, the amyloid cascade may be a consequence of earlier abnormalities in multiple neurobiological pathways; that is, it may be a cascade within a cascade.

The cluster analyses of signal intensity profiles in 70 transcripts suggest the possibility that active or dynamic protective responses occur in the BA 1/2/3, as presumptive susceptibility related changes in expression are occurring in BA 21. Nevertheless, it is possible that these changes are obligatory adaptations to E4 and are neither protective nor pathogenic in nature, that is, are neutral. Although we cannot fully rule out the latter hypothesis, we do not believe that is it likely because many of the individual genes and pathways have been implicated in AD. We also appreciate that not every individual with an APOE4 allele will develop AD. Nevertheless, as based on population prevalence figures, we estimate that only about two to three APOE3 individuals (of 28), while five to six APOE4 individuals (of 13) in the sample would have gone on to manifest AD.

The fold-change differences that we observed were often subtle. However, they survived rigorous statistical procedures. Moreover, because our subjects died B30 years on average before any of the clinical manifestations of dementia, we expected that fold-change differences would be rather small, but nevertheless measurable and reproducible.

Several previous human microarray studies with emphasis on APOE or preclinical AD are relevant to our study. Although these studies did not directly examine regional differences, nor use APOE status in younger cases to separate early pathogenic processes from the consequences of pathology, there are nevertheless several points of comparison, including abnormalities in the expression of such individual transcripts as NDUFB7 and GRIA2, transcription-repressor factor upregulation and mitochondrial dysfunction.61-65 Detailed discussion of this human post-mortem study, as well as alternative cellular approaches to vulnerability, can be found in the Supplementary Discussion section.

More broadly, the long molecular prodrome that we found in APOE4-related AD is consistent with several neurophysiological studies and some neurocognitive studies. Both glucose metabolism reductions and abnormalities in fMRI BOLD signal in temporal lobe structures of APOE4 carriers have been found.66,67 APOE4 may also negatively affect memory function and rate of decline in younger populations.68

To summarize, we found that APOE variants differed in multiple transcripts and multiple biological pathways, including those affecting calcium regulation, cell-cycle reentry and apoptosis, mitochondrial function and transcription factors. This may be consistent with broad notions about why the differences in APOE isoforms can ‘translate up’ to a devastating neurodegenerative disease,69 as several pathways may interact and ultimately promote amyloid aggregation or tau fibrillization.

Supplementary Material

Supp File

Acknowledgments

We thank Dr Richard Simon for providing expert statistical advice and Ms Amy Deep-Soboslay and Dr Llewellyn Bigelow for post-mortem patient screening and diagnosis. We thank Mr Brady Kirchberg for providing expertise in graphical display. We also thank Dr Franak Batliwalla and Ms Aarti Damle, members of the Feinstein Institute’s microarray core facility, for their assistance.

Footnotes

Conflict of interest TG has consulted for Merck and GSK. He receives royalties for use of a cognitive test battery in clinical trials, the BACS. He has received an investigator initiated grant from Eisai/Pfizer. PD has received research support from and served as a consultant to Applied Neurosolutions. The remaining authors declare no conflict of interest.

Supplementary Information accompanies the paper on the Molecular Psychiatry website (http://www.nature.com/mp)

References

  • 1.Braak H, Del Tredici K, Schultz C, Braak E. Vulnerability of select neuronal types to Alzheimer’s disease. Ann NY Acad Sci. 2000;924:53–61. doi: 10.1111/j.1749-6632.2000.tb05560.x. [DOI] [PubMed] [Google Scholar]
  • 2.Haroutunian V, Katsel P, Schmeidler J. Transcriptional vulnerability of brain regions in Alzheimer’s disease and dementia. Neurobiol Aging. 2007;30:561–573. doi: 10.1016/j.neurobiolaging.2007.07.021. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 3.Arnold SE, Hyman BT, Flory J, Damasio AR, Van Hoesen GW. The topographical and neuroanatomical distribution of neurofibrillary tangles and neuritic plaques in the cerebral cortex of patients with Alzheimer’s disease. Cerebral Cortex. 1991;1:103–116. doi: 10.1093/cercor/1.1.103. [DOI] [PubMed] [Google Scholar]
  • 4.Brun A, Gustafson L. Distribution of cerebral degeneration in Alzheimer’s disease. A clinico-pathological study. Archiv fur Psychiatrie und Nervenkrankheiten. 1976;223:15–33. doi: 10.1007/BF00367450. [DOI] [PubMed] [Google Scholar]
  • 5.Coon KD, Myers AJ, Craig DW, Webster JA, Pearson JV, Lince DH, et al. A high-density whole-genome association study reveals that APOE is the major susceptibility gene for sporadic late-onset Alzheimer’s disease. J Clin Psychiatry. 2007;68:613–618. doi: 10.4088/jcp.v68n0419. [see comment] [DOI] [PubMed] [Google Scholar]
  • 6.Reddy PH, McWeeney S. Mapping cellular transcriptosomes in autopsied Alzheimer’s disease subjects and relevant animal models. Neurobiol Aging. 2006;27:1060–1077. doi: 10.1016/j.neurobiolaging.2005.04.014. [see comment] [DOI] [PubMed] [Google Scholar]
  • 7.Katsel PL, Davis KL, Haroutunian V. Large-scale microarray studies of gene expression in multiple regions of the brain in schizophrenia and Alzheimer’s disease. Int Rev Neurobiol. 2005;63:41–82. doi: 10.1016/S0074-7742(05)63003-6. [DOI] [PubMed] [Google Scholar]
  • 8.Harrison PJ, Heath PR, Eastwood SL, Burnet PW, McDonald B, Pearson RC. The relative importance of premortem acidosis and postmortem interval for human brain gene expression studies: selective mRNA vulnerability and comparison with their encoded proteins. Neurosci Lett. 1995;200:151–154. doi: 10.1016/0304-3940(95)12102-a. [DOI] [PubMed] [Google Scholar]
  • 9.Mirnics K, Levitt P, Lewis DA. DNA microarray analysis of postmortem brain tissue. Int Rev Neurobiol. 2004;60:153–181. doi: 10.1016/S0074-7742(04)60006-7. [DOI] [PubMed] [Google Scholar]
  • 10.Conejero-Goldberg C, Wang E, Yi C, Goldberg TE, Jones-Brando L, Marincola FM, et al. Infectious pathogen detection arrays: viral detection in cell lines and postmortem brain tissue. Biotechniques. 2005;39:741–751. doi: 10.2144/000112016. [DOI] [PubMed] [Google Scholar]
  • 11.Khachaturian ZS. Diagnosis of Alzheimer’s disease. Arch Neurol. 1985;42:1097–1105. doi: 10.1001/archneur.1985.04060100083029. [DOI] [PubMed] [Google Scholar]
  • 12.Hixson JE, Vernier DT. Restriction isotyping of human apolipoprotein E by gene amplification and cleavage with HhaI. J Lipid Res. 1990;31:545–548. [PubMed] [Google Scholar]
  • 13.Barnes M, Freudenberg J, Thompson S, Aronow B, Pavlidis P. Experimental comparison and cross-validation of the Affymetrix and Illumina gene expression analysis platforms. Nucleic Acids Res. 2005;33:5914–5923. doi: 10.1093/nar/gki890. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.McShane LM, Radmacher MD, Freidlin B, Yu R, Li MC, Simon R. Methods for assessing reproducibility of clustering patterns observed in analyses of microarray data. Bioinformatics. 2002;18:1462–1469. doi: 10.1093/bioinformatics/18.11.1462. [DOI] [PubMed] [Google Scholar]
  • 15.Calvano SE, Xiao W, Richards DR, Felciano RM, Baker HV, Cho RJ, et al. A network-based analysis of systemic inflammation in humans. Nature. 2005;437:1032–1037. doi: 10.1038/nature03985. [published erratum appears in Nature 2005;438: 696] [DOI] [PubMed] [Google Scholar]
  • 16.LeBlanc AC. The role of apoptotic pathways in Alzheimer’s disease neurodegeneration and cell death. Curr Alzheimer Res. 2005;2:389–402. doi: 10.2174/156720505774330573. [DOI] [PubMed] [Google Scholar]
  • 17.Nikolaev A, McLaughlin T, O’Leary DD, Tessier-Lavigne M. APP binds DR6 to trigger axon pruning and neuron death via distinct caspases. Nature. 2009;457:981–989. doi: 10.1038/nature07767. [see comment] [DOI] [PMC free article] [PubMed] [Google Scholar] [Retracted]
  • 18.Klaiman G, Petzke TL, Hammond J, Leblanc AC. Targets of caspase-6 activity in human neurons and Alzheimer’s disease. Mol Cell Proteomics. 2008;7:1541–1555. doi: 10.1074/mcp.M800007-MCP200. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 19.Yasuda RP, Ikonomovic MD, Sheffield R, Rubin RT, Wolfe BB, Armstrong DM. Reduction of AMPA-selective glutamate receptor subunits in the entorhinal cortex of patients with Alzheimer’s disease pathology: a biochemical study. Brain Res. 1995;678:161–167. doi: 10.1016/0006-8993(95)00178-s. [DOI] [PubMed] [Google Scholar]
  • 20.Carter TL, Rissman RA, Mishizen-Eberz AJ, Wolfe BB, Hamilton RL, Gandy S, et al. Differential preservation of AMPA receptor subunits in the hippocampi of Alzheimer’s disease patients according to Braak stage. Exp Neurol. 2004;187:299–309. doi: 10.1016/j.expneurol.2003.12.010. [DOI] [PubMed] [Google Scholar]
  • 21.Altar CA, Vawter MP, Ginsberg SD. Target identification for CNS diseases by transcriptional profiling. Neuropsychopharmacol Rev. 2009;34:18–54. doi: 10.1038/npp.2008.172. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 22.Zhang YW, Wang R, Liu Q, Zhang H, Liao FF, Xu H. Presenilin/gamma-secretase-dependent processing of beta-amyloid precursor protein regulates EGF receptor expression. Proc Natl Acad Sci USA. 2007;104:10613–10618. doi: 10.1073/pnas.0703903104. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 23.Repetto E, Yoon IS, Zheng H, Kang DE. Presenilin 1 regulates epidermal growth factor receptor turnover and signaling in the endosomal-lysosomal pathway. J Biol Chem. 2007;282:31504–31516. doi: 10.1074/jbc.M704273200. [DOI] [PubMed] [Google Scholar]
  • 24.Cha YK, Kim YH, Ahn YH, Koh JY. Epidermal growth factor induces oxidative neuronal injury in cortical culture. J Neurochem. 2000;75:298–303. doi: 10.1046/j.1471-4159.2000.0750298.x. [DOI] [PubMed] [Google Scholar]
  • 25.Kang CB, Hong Y, Dhe-Paganon S, Yoon HS. FKBP family proteins: immunophilins with versatile biological functions. Neurosignals. 2008;16:318–325. doi: 10.1159/000123041. [DOI] [PubMed] [Google Scholar]
  • 26.Hoeffer CA, Tang W, Wong H, Santillan A, Patterson RJ, Martinez LA, et al. Removal of FKBP12 enhances mTOR-Raptor interactions, LTP, memory, and perseverative/repetitive behavior. Neuron. 2008;60:832–845. doi: 10.1016/j.neuron.2008.09.037. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 27.Wang HQ, Nakaya Y, Du Z, Yamane T, Shirane M, Kudo T, et al. Interaction of presenilins with FKBP38 promotes apoptosis by reducing mitochondrial Bcl-2. Hum Mol Genet. 2005;14:1889–1902. doi: 10.1093/hmg/ddi195. [DOI] [PubMed] [Google Scholar]
  • 28.Gutman CR, Strittmatter WJ, Weisgraber KH, Matthew WD. Apolipoprotein E binds to and potentiates the biological activity of ciliary neurotrophic factor. J Neurosci. 1997;17:6114–6121. doi: 10.1523/JNEUROSCI.17-16-06114.1997. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 29.Qu HY, Zhang T, Li XL, Zhou JP, Zhao BQ, Li Q, et al. Transducible P11-CNTF rescues the learning and memory impairments induced by amyloid-beta peptide in mice. Eur J Pharmacol. 2008;594:93–100. doi: 10.1016/j.ejphar.2008.06.109. [DOI] [PubMed] [Google Scholar]
  • 30.Laity JH, Lee BM, Wright PE. Zinc finger proteins: new insights into structural and functional diversity. Curr Opin Struct Biol. 2001;11:39–46. doi: 10.1016/s0959-440x(00)00167-6. [DOI] [PubMed] [Google Scholar]
  • 31.Maguschak KA, Ressler KJ. Beta-catenin is required for memory consolidation. Nature Neurosci. 2008;11:1319–1326. doi: 10.1038/nn.2198. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 32.Sasaki T, Hirabayashi J, Manya H, Kasai K, Endo T. Galectin-1 induces astrocyte differentiation, which leads to production of brain-derived neurotrophic factor. Glycobiology. 2004;14:357–363. doi: 10.1093/glycob/cwh043. [DOI] [PubMed] [Google Scholar]
  • 33.Cooper D, Norling LV, Perretti M. Novel insights into the inhibitory effects of Galectin-1 on neutrophil recruitment under flow. J Leukoc Biol. 2008;83:1459–1466. doi: 10.1189/jlb.1207831. [DOI] [PubMed] [Google Scholar]
  • 34.La M, Cao TV, Cerchiaro G, Chilton K, Hirabayashi J, Kasai K, et al. A novel biological activity for galectin-1: inhibition of leukocyte-endothelial cell interactions in experimental inflammation. Am J Pathol. 2003;163:1505–1515. doi: 10.1016/s0002-9440(10)63507-9. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 35.Matsuda A, Suzuki Y, Honda G, Muramatsu S, Matsuzaki O, Nagano Y, et al. Large-scale identification and characterization of human genes that activate NF-kappaB and MAPK signaling pathways. Oncogene. 2003;22:3307–3318. doi: 10.1038/sj.onc.1206406. [DOI] [PubMed] [Google Scholar]
  • 36.Zhang JY, Liu SJ, Li HL, Wang JZ. Microtubule-associated protein tau is a substrate of ATP/Mg(2 + )-dependent proteasome protease system. J Neural Transm. 2005;112:547–555. doi: 10.1007/s00702-004-0196-x. [DOI] [PubMed] [Google Scholar]
  • 37.Salon M Lopez, Pasquini L, Moreno M Besio, Pasquini JM, Soto E. Relationship between beta-amyloid degradation and the 26S proteasome in neural cells. Exp Neurol. 2003;180:131–143. doi: 10.1016/s0014-4886(02)00060-2. [DOI] [PubMed] [Google Scholar]
  • 38.Cecarini V, Bonfili L, Amici M, Angeletti M, Keller JN, Eleuteri AM. Amyloid peptides in different assembly states and related effects on isolated and cellular proteasomes. Brain Res. 2008;1209:8–18. doi: 10.1016/j.brainres.2008.03.003. [DOI] [PubMed] [Google Scholar]
  • 39.Hoglinger GU, Carrard G, Michel PP, Medja F, Lombes A, Ruberg M, et al. Dysfunction of mitochondrial complex I and the proteasome: interactions between two biochemical deficits in a cellular model of Parkinson’s disease. J Neurochem. 2003;86:1297–1307. doi: 10.1046/j.1471-4159.2003.01952.x. [DOI] [PubMed] [Google Scholar]
  • 40.Stutzmann GE. The pathogenesis of Alzheimer’s disease is it a lifelong ‘calciumopathy’? Neuroscientist. 2007;13:546–559. doi: 10.1177/1073858407299730. [DOI] [PubMed] [Google Scholar]
  • 41.LaFerla FM. Calcium dyshomeostasis and intracellular signalling in Alzheimer’s disease. Nat Rev Neurosci. 2002;3:862–872. doi: 10.1038/nrn960. [DOI] [PubMed] [Google Scholar]
  • 42.Dreses-Werringloer U, Lambert JC, Vingtdeux V, Zhao H, Vais H, Siebert A, et al. A polymorphism in CALHM1 influences Ca2 homeostasis, Abeta levels, and Alzheimer’s disease risk. Cell. 2008;133:1149–1161. doi: 10.1016/j.cell.2008.05.048. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 43.Mudher A, Lovestone S. Alzheimer’s disease-do tauists and baptists finally shake hands? Trends Neurosci. 2002;25:22–26. doi: 10.1016/s0166-2236(00)02031-2. [DOI] [PubMed] [Google Scholar]
  • 44.De Strooper B, Annaert W. Where Notch and Wnt signaling meet. The presenilin hub. J Cell Biol. 2001;152:785–794. doi: 10.1083/jcb.152.4.f17. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 45.Andorfer C, Acker CM, Kress Y, Hof PR, Duff K, Davies P. Cell-cycle reentry and cell death in transgenic mice expressing nonmutant human tau isoforms. J Neurosci. 2005;25:5446–5454. doi: 10.1523/JNEUROSCI.4637-04.2005. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 46.Herrup K, Arendt T. Re-expression of cell cycle proteins induces neuronal cell death during Alzheimer’s disease. J Alzheimers Dis. 2002;4:243–247. doi: 10.3233/jad-2002-4315. [see comment] [DOI] [PubMed] [Google Scholar]
  • 47.Herrup K, Neve R, Ackerman SL, Copani A. Divide and die: cell cycle events as triggers of nerve cell death. J Neurosci. 2004;24:9232–9239. doi: 10.1523/JNEUROSCI.3347-04.2004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 48.Yang Y, Mufson EJ, Herrup K. Neuronal cell death is preceded by cell cycle events at all stages of Alzheimer’s disease. J Neurosci. 2003;23:2557–2563. doi: 10.1523/JNEUROSCI.23-07-02557.2003. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 49.Mattson MP, Keller JN, Begley JG. Evidence for synaptic apoptosis. Exp Neurol. 1998;153:35–48. doi: 10.1006/exnr.1998.6863. [DOI] [PubMed] [Google Scholar]
  • 50.Albrecht S, Bourdeau M, Bennett D, Mufson EJ, Bhattacharjee M, LeBlanc AC. Activation of caspase-6 in aging and mild cognitive impairment. Am J Pathol. 2007;170:1200–1209. doi: 10.2353/ajpath.2007.060974. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 51.Zhong N, Scearce-Levie K, Ramaswamy G, Weisgraber KH. Apolipoprotein E4 domain interaction: synaptic and cognitive deficits in mice. Alzheimers Dement. 2008;4:179–192. doi: 10.1016/j.jalz.2008.01.006. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 52.Mahley RW, Huang Y. Apolipoprotein (apo) E4 and Alzheimer’s disease: unique conformational and biophysical properties of apoE4 can modulate neuropathology. Acta Neurologica Scandinavica Supplementum. 2006;185:8–14. doi: 10.1111/j.1600-0404.2006.00679.x. [DOI] [PubMed] [Google Scholar]
  • 53.Reddy PH, Beal MF. Are mitochondria critical in the pathogenesis of Alzheimer’s disease? Brain Res Rev. 2005;49:618–632. doi: 10.1016/j.brainresrev.2005.03.004. [DOI] [PubMed] [Google Scholar]
  • 54.Lin MT, Beal MF. Mitochondrial dysfunction and oxidative stress in neurodegenerative diseases. Nature. 2006;443:787–795. doi: 10.1038/nature05292. [DOI] [PubMed] [Google Scholar]
  • 55.Starkov AA, Beal FM. Portal to Alzheimer’s disease. Nat Med. 2008;14:1020–1021. doi: 10.1038/nm1008-1020. [comment] [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 56.Smith MA, Drew KL, Nunomura A, Takeda A, Hirai K, Zhu X, et al. Amyloid-beta, tau alterations and mitochondrial dysfunction in Alzheimer disease: the chickens or the eggs? Neurochem Int. 2002;40:527–531. doi: 10.1016/s0197-0186(01)00123-1. [DOI] [PubMed] [Google Scholar]
  • 57.Andersen JK. Oxidative stress in neurodegeneration: cause or consequence? Nat Med. 2004;10(Suppl):S18–S25. doi: 10.1038/nrn1434. [DOI] [PubMed] [Google Scholar]
  • 58.Liang WS, Reiman EM, Valla J, Dunckley T, Beach TG, Grover A, et al. Alzheimer’s disease is associated with reduced expression of energy metabolism genes in posterior cingulate neurons. Proc Natl Acad Sci USA. 2008;105:4441–4446. doi: 10.1073/pnas.0709259105. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 59.Rhein V, Song X, Wiesner A, Ittner LM, Baysang G, Meier F, et al. Amyloid-beta and tau synergistically impair the oxidative phosphorylation system in triple transgenic Alzheimer’s disease mice. Pro Natl Acad Sci USA. 2009;106:20057–20062. doi: 10.1073/pnas.0905529106. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 60.Hardy J, Selkoe DJ. The amyloid hypothesis of Alzheimer’s disease: progress and problems on the road to therapeutics. Science. 2002;297:353–356. doi: 10.1126/science.1072994. [see comment]. [published erratum appears in Science 2002; 297: 2209] [DOI] [PubMed] [Google Scholar]
  • 61.Xu PT, Li YJ, Qin XJ, Scherzer CR, Xu H, Schmechel DE, et al. Differences in apolipoprotein E3/3 and E4/4 allele-specific gene expression in hippocampus in Alzheimer disease. Neurobiol Dis. 2006;21:256–275. doi: 10.1016/j.nbd.2005.07.004. [DOI] [PubMed] [Google Scholar]
  • 62.Liang WS, Dunckley T, Beach TG, Grover A, Mastroeni D, Ramsey K, et al. Altered neuronal gene expression in brain regions differentially affected by Alzheimer’s disease: a reference data set. Physiol Genom. 2008;33:240–256. doi: 10.1152/physiolgenomics.00242.2007. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 63.Blalock EM, Geddes JW, Chen KC, Porter NM, Markesbery WR, Landfield PW. Incipient Alzheimer’s disease: microarray correlation analyses reveal major transcriptional and tumor suppressor responses. Proc Natl Acad Sci USA. 2004;101:2173–2178. doi: 10.1073/pnas.0308512100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 64.Ginsberg SD, Hemby SE, Lee VM, Eberwine JH, Trojanowski JQ. Expression profile of transcripts in Alzheimer’s disease tangle-bearing CA1 neurons. Ann Neurol. 2000;48:77–87. [PubMed] [Google Scholar]
  • 65.Dunckley T, Beach TG, Ramsey KE, Grover A, Mastroeni D, Walker DG, et al. Gene expression correlates of neurofibrillary tangles in Alzheimer’s disease. Neurobiol Aging. 2006;27:1359–1371. doi: 10.1016/j.neurobiolaging.2005.08.013. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 66.Reiman EM, Chen K, Alexander GE, Caselli RJ, Bandy D, Osborne D, et al. Functional brain abnormalities in young adults at genetic risk for late-onset Alzheimer’s dementia. Proc Natl Acad Sci USA. 2004;101:284–289. doi: 10.1073/pnas.2635903100. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 67.Bookheimer SY, Strojwas MH, Cohen MS, Saunders AM, Pericak-Vance MA, Mazziotta JC, et al. Patterns of brain activation in people at risk for Alzheimer’s disease. N Engl J Med. 2000;343:450–456. doi: 10.1056/NEJM200008173430701. [see comment] [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 68.Caselli RJ, Dueck AC, Osborne D, Sabbagh MN, Connor DJ, Ahern GL, et al. Longitudinal modeling of age-related memory decline and the APOE epsilon4 effect. N Engl J Med. 2009;361:255–263. doi: 10.1056/NEJMoa0809437. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 69.Palop JJ, Chin J, Mucke L. A network dysfunction perspective on neurodegenerative diseases. Nature. 2006;443:768–773. doi: 10.1038/nature05289. [DOI] [PubMed] [Google Scholar]

Associated Data

This section collects any data citations, data availability statements, or supplementary materials included in this article.

Supplementary Materials

Supp File
NIHMS212621-supplement-Supp_File.pdf (482.4KB, pdf)

RESOURCES