|Home | About | Journals | Submit | Contact Us | Français|
APOE4 allele is a major risk factor for late-onset Alzheimer disease (AD). The mechanism of action of APOE in AD remains unclear. To study the effects of APOE alleles on gene expression in AD, we have analyzed the gene transcription patterns of human hippocampus from APOE3/3, APOE3/4, APOE4/4 AD patients and normal control using Serial Analysis of Gene Expression (SAGE). Using SAGE, we found gene expression patterns in hippocampus of APOE3/4 and APOE4/4 AD patients differ substantially from those of APOE3/3 AD patients. APOE3/4 and APOE4/4 allele expression may activate similar genes or gene pools with associated functions. APOE4 AD alleles activate multiple tumor suppressors, tumor inducers and negative regulator of cell growth or repressors that may lead to increased cell arrest, senescence and apoptosis. In contrast, there is decreased expression of large clusters of genes associated with synaptic plasticity, synaptic vesicle docking and fusing and axonal/neuronal outgrowth. In addition, reduction of neurotransmitter receptors and Ca++ homeostasis, disruption of multiple signal transduction pathways, loss of cell protection, and perhaps most notably, mitochondrial oxidative phosphorylation/energy metabolism are associated with APOE3/4 and APOE4/4 AD alleles. These findings may help define the mechanisms that APOE4 contribute increased risk for AD and identify new candidate genes conferring susceptibility to AD.
Alzheimer disease (AD) is a neurodegenerative disorder characterized by progressive deterioration of memory and cognition with specific neuropathological changes, including extracellular amyloid –containing plaques, intracellular neurofibrillary tangles (NFT) of abnormally phosphorylated tau protein, and degeneration of the cholinergic neurons of the basal forebrain (Auld et al., 2002). A relatively small number of early onset AD families exhibit genetic linkage to mutations in presenilin-1 (PS1), presenilin-2 (PS2), and amyloid precursor protein (APP) genes (George-Hyslop et al., 2000;Tamouza et al., 2000;Holmes et al., 2000). The three major isoforms of human apolipoprotein E (apoE2, E3 and E4) are associated with both risk and age of onset of AD and response to brain injury. The APOE4 allele is associated with increased risk and earlier age of onset; while the APOE2 allele decreases risk and delays AD onset in population studies compared to the APOE3 allele. (Corder et al., 1993;Saunders et al., 1993;Poirier et al., 1993;Rebeck et al., 1993). The APOE4 allele increases risk and lowers the age of onset of AD in a dose-dependent manner (Corder et al., 1994). In addition, the APOE4 allele has been implicated in poorer neurological recovery from head injury, cerebral hemorrhage, and cognitive status after cardiac bypass surgery (Alberts et al., 1995;Tardiff et al., 1997). The exact role that apoE plays in neuronal metabolism and function is, at present, poorly understood.
Several microarray studies involving human AD brain have been published (Eikelboom et al., 2000;Auld et al., 2002;Mufson et al., 2002;Colangelo et al., 2002;Beckmann et al., 2001;Pasinetti and Ho, 2001;Blalock et al., 2004). These studies have yielded important new insights regarding changes in gene transcription in AD brain. This laboratory recently extended microarray studies by comparing APOE3/3 and APOE4/4 allele-specific microarray gene expression profiles from hippocampus of AD cases (Xu et al., 2006).
Serial analysis of gene expression (SAGE) is a sequencing-based technique used to quantify the relative expression levels of thousands of transcripts by sequencing concatemers of short sequence tags (10 bp +restriction site) derived from biological samples (Velculescu et al., 1995). This technique has several important advantages compared to other techniques for detection of tissue specific expression (Evans et al., 2002). First, SAGE creates a permanent, quantitative record of the set of sequences transcribed in a given tissue or cell population making it suited for the quantitative analysis of gene transcripts and detection of new and/or novel genes associated with diseases, such as AD. Second, SAGE can detect small changes in expression levels (Scott and Chrast, 2001). Third, transcripts that are over or under-expressed can be detected equally well (Iyer and Struhl, 1996). The SAGE quantification profiling of AD based on APOE genotypes (alleles) has not been documented in previous studies.
We now extend our gene transcription studies by generating and analyzing APOE allele specific human SAGE libraries derived from hippocampus from AD patients. The hippocampus is heavily involved in the neuropathology of AD and is often selected for both neuropathological and molecular analysis (Markesbery, 1997). In this study, we selected hippocampus samples from AD cases with APOE4/4, APOE3/4, and APOE3/3 genotypes and controls of APOE 3/3 genotype.
We generated four SAGE libraries using human brain hippocampus RNA isolated from AD patients carrying APOE3/3, APOE3/4 and APOE4/4 alleles, and a normal control (APOE3/3). A total of 288,661 tags, with an average of 72,165 tags for each library, were extracted from sequence files from these four SAGE libraries using e-SAGE software (Table 2). These tags correspond to 95,096 (32.9 %) tags of different transcripts, with an average of 23,774 unique tags per library. These unique tags are mapped to 54,292 unique UniGene clusters. Comparison of APOE allele-specific AD SAGE libraries with control library and between AD SAGE libraries indicated remarkable similarity in expression patterns or profiles (Figure 1). The correlation coefficients were in the range of 0.89 to 0.99. The eSAGE database comparisons revealed similar numbers of total tags compared between each pair (Table 3, Fig. 2A). In APOE3/4 AD and APOE4/4 AD compared with both APOE3/3 AD and control, larger number of tags (625-918) showed changes greater than two-fold (P < 0.05) than tags (156) in the APOE3/3 AD vs. control (Table 3, Fig. 2B). Similarly, there are more up- and down-regulated total mapped UniGene clusters (Fig 2C) and unique UniGene clusters (Fig. 2D) in APOE3/4 AD and APOE4/4 AD vs. APOE3/3 AD and APOE3/3 control than those in APOE3/3 AD vs. control. Less than one quarter (18.8-21.2 %) of the tags did not map to any UniGene cluster (orphan tags) (Table 2). These results are consistent with previously reported SAGE libraries (Gunnersen et al., 2002;Scott and Chrast, 2001;Anisimov et al., 2002).
To explore whether specific APOE alleles may regulate or affect different genes in AD, we compared differentially expressed genes in APOE3/3 AD, APOE3/4 AD and APOE4/4 AD versus APOE3/3 normal controls. Comparing APOE3/3 AD and APOE3/3 control SAGE libraries, the distribution of SAGE tag counts are quite similar with most of the SAGE tags lying close to the 45-degree identity line (Figure 1A). A total of 37,900 tags were compared and 156 tags (0.41 %) were found with significant tag count differences (P< 0.05). One hundred tags could be specifically assigned to unique UniGene Clusters (Table 3). Of this total, forty-eight gene transcripts are up-regulated and 52 genes down-regulated (Table 3, Fig. 2D). For APOE3/4 AD and APOE4/4 AD vs. control, 39,059 and 36,632 SAGE tags were compared and 906 tags (2.32 %) and 625 tags (1.71 %) showed significant results, respectively. 523 and 320 tags were assigned to unique UniGene Clusters and 97 and 108 UniGene clusters are up-regulated and 426 and 212 UniGene clusters down-regulated, respectively (Fig. 1B,1C, Table 3, Fig. 2D). Direct comparison between the three AD SAGE libraries and control reveal APOE allele-specific quantitative differences. In general more total gene transcripts are up-regulated and down-regulated in both APOE3/4 AD and APOE4/4 AD than in APOE3/3 AD hippocampus relative to control (Table 3; Fig. 1, ,22).
Among the upregulated genes in APOE4/4 AD and APOE3/4 AD versus control, 15 out of 108 (13.9 %) and 97 (15.5 %) up-regulated transcripts overlapped (Fig 5A). Only 4 genes (3.7 %) in APOE4/4 AD vs. control and 3 genes (3.1 %) in APOE3/4 AD vs. control shared upregulated genes in APOE3/3 AD vs. control (Fig. 5A). Among the 212 downregulated gene transcripts in APOE4/4 AD versus control, 135 genes (63.7 %) were also found downregulated in 426 genes in the comparison of APOE3/4 AD vs. control (Fig 5B). In contrast, only 16 genes (7.5 %) in the APOE4/4 AD vs. control and 28 genes (6.6 %) in the APOE3/4 AD vs. control were shared downregulated genes in APOE3/3 AD vs. control (Fig. 5B). Detail overlapped genes and fold changes were listed in Supplementary Table 1-3 online. These findings support the possibility that APOE4 gene expression may regulate different genes or gene pools from the APOE3 gene expression in hippocampus of AD patients.
A total of 918 and 771 SAGE tags are significantly (p<0.05) differentially expressed in APOE3/4 AD vs. APOE3/3 AD and APOE4/4 AD vs. APOE3/3 AD, respectively. These tags were specifically assigned to 499 and 428 unique UniGene clusters with 131 and 172 gene transcripts up-regulated and 368 and 256 genes down-regulated in these two comparisons (Table 3, Fig. 2D). Among up-regulated gene transcripts, 20 genes out of 131 transcripts (15.3 %) in APOE3/4 AD vs. APOE3/3 AD and 172 (11.6 %) genes in APOE4/4 AD vs. APOE3/3 AD are shared up-regulated genes in these two comparisons (Fig. 6A, Supplementary table 4). In contrast, 167 out of 368 (45 %) and 256 (65 %) are shared down-regulated genes in these two comparisons (Fig 6B). These observations suggest that APOE4 allele gene expression in hippocampus of AD may down-regulate or inactivate a subset of common genes and/or the same gene pools with similar or associated functions.
Among the upregulated genes in the hippocampus of APOE3/4 and APOE4/4 AD compared to APOE3/3 AD and APOE3/3 control, 66 and 58 genes shared upregulation in both comparisons (Fig.7A, Supplementary Table 7, 8). Among the down-regulated genes in the hippocampus of APOE3/4 AD and APOE4/4 AD compared to APOE3/3 AD and APOE3/3 control, 219 and 120 downregulated gene transcripts overlapped (Fig. 7B, Supplementary Table 7, 8).
Molecular functional categorization analysis (EASE) demonstrates that transcripts of protein binding and transporter, receptor binding and signaling, nucleotide binding, transferase and kinase activity are among the most elevated gene categories in both APOE3/4 AD vs. control and APOE4/4 AD vs. control compared to APOE3/3 AD vs. control (Fig. 3A, Table 4). In APOE3/4 AD vs. control genes involved in protein binding and transporter, receptor binding and signaling, carriers and enzyme activity are the most up-regulated transcripts relative to both APOE3/3 and APOE4/4 AD compared with control (Fig. 3A Table 4). In APOE4/4 AD hippocampus, larger numbers of transcripts encoding hydrolase, electron/channel transporter, kinase and oxidoreductase are up-regulated compared to APOE3/3 AD and APOE3/4 AD.
A large number of gene transcripts are down-regulated in APOE3/4 and APOE4/4 AD relative to APOE3/3 AD when compared to control in 14 selected functional categories (Fig. 3B, Table 4). The APOE3/4 AD SAGE library has the most transcripts down-regulated in 13 of the 14 selected functional categories of genes. Theses findings are consistent with recent published microarray gene expression studies (Xu et al., 2006) that APOE4 gene expression in hippocampus may activate or inhibit multiple categories of protein function and/or protein expression compared with APOE3 expression in different AD patients.
To investigate whether there are fundamental differences in gene expression in AD patients carrying one, two, or no copies of the APOE4 allele the molecular functional categories of gene transcripts between APOE3/4 vs. APOE3/3 and APOE4/4 vs. APOE3/3 AD SAGE libraries were compared (Fig. 4, Table 5). Three functional categories of transcripts (hydrolase, nucleic acid binding and transferase) have more upregulated transcripts in APOE4/4 AD than APOE3/4 AD hippocampus (Fig. 4A). Several functional categories have larger numbers of elevated transcripts in APOE4/4 AD vs. APOE3/4 AD than in APOE4/4 AD vs. APOE3/3 AD and APOE3/4 AD vs. APOE3/3 AD. Categories include protein binding and transport, receptor binding and signaling, nucleotide binding, metal ion binding, hydrolase and kinase (Fig. 4A). In contrast, there are fewer significant down-regulated gene transcripts in all 14 selected functional categories identified in the comparison of APOE4/4 AD vs. APOE3/4 AD than in the comparisons of APOE4/4 AD vs. APOE3/3 AD and APOE3/4 AD vs. APOE3/3 AD (Fig. 4B, Table 5).
To examine genes whose transcription is substantially altered we chose five fold transcription alteration as the cutoff and directly compared major functional categories that are 5-fold or greater up- or down-regulated genes in APOE3/3, APOE3/4, APOE4/4 AD tissues compared to APOE3/3 control (Tables 7--11;11; Supplementary Table 9-11 online). One striking difference is that APOE3/4 and APOE4/4 alleles strongly up-regulate gene transcription of tumor suppressor, cell arrest and negative regulators of cell growth and repressors (Table 7). Other clear differences are up-regulation of genes associated with protein modification and oxidoreductase. Several gene clusters are up-regulated in all three AD genotypes, including calcium signaling and homeostasis, cell adhesion, myelinogenesis, transcription factors and regulators, cell growth and differentiation (Table 7).
Functional clusters containing the greatest number of 5-fold down-regulated transcripts in APOE3/4 AD are genes involved in regulating signal transduction pathways (including G-protein mediated signaling), synaptic plasticity and formation, synaptic vesicle docking and fusion, and axonal guidance, neuronal outgrowth and neurogenesis (Table 9, ,10).10). Although individual calcium signaling transcripts are up-regulated (Table 7) in all three APOE genotypes of ADs vs. control, signal transduction pathways in general are down-regulated especially in APOE3/4 AD and APOE4/4 AD compared with control (Table 10, Supplementary Table 10, 11).
Other groups of down-regulated gene transcripts include proteins that regulate cell growth, differentiation and development, and protein biosynthesis. Apoptosis inducers are down-regulated especially in APOE3/4 AD hippocampus compared with APOE3/3 control (Supplementary Table 10). Negative regulators or repressors and genes involved in negative growth regulation are down-regulated in APOE4/4 AD and APOE3/4 AD compared to control (Supplementary Table 10, 11). There are more down-regulated gene transcripts in APOE3/4 AD than in APOE4/4 AD that involves protein modification, ubiquitination, ER to Golgi transport and vesicle trafficking and RNA splicing and RNA editing have (Supplementary Table 9-11). Consistent with previously reported results, multiple gene transcripts encoding neurotransmitter receptors, energy metabolism and mitochondrial oxidative phosphorylation and cytoskeleton and structure proteins are significantly down-regulated in both APOE3/4 AD and APOE4/4 AD hippocampus relative to APOE3/3 control (Table 11, Supplementary Table 9-11).
Comparing APOE3/4 and APOE4/4 AD with APOE3/3 AD and control, there are similar patterns of the functional clustering of highly up-regulated (Table 7, ,8)8) and down-regulated genes (Table 9--11;11; Supplementary Table 12, 13). Transcripts associated with signaling transduction are among the most elevated category in the both APOE3/4 AD and APOE4/4 AD (Table 8). Tumor suppressor and cell arrest gene transcripts are up-regulated in both APOE3/4 AD and APOE4/4 AD compared with APOE3/3 AD (Table 8), as well as APOE3/3 control (Table 7). In APOE3/4 AD cases compared with APOE3/3 AD, there are more up-regulated transcripts encoding cell adhesion and myelinogenesis, apoptosis, ER to Golgi transport and vesicle trafficking, and stress response and chaperone (Table 8, Supplementary Table 12, 13). In the APOE4/4 AD vs. APOE3/3 AD, there are more upregulated transcripts in the functional categories of cell growth and differentiation, protein modification, RNA binding and editing.
In APOE3/4 AD compared with APOE3/3 AD, signal transduction, calcium signaling and homeostasis, synaptic plasticity, synaptic vesicle docking and fusing, neurite outgrowth, RNA processing and editing, cell adhesion and myelinogenesis are the most down-regulated transcript clusters (Supplementary Table 12). In the functional categories of negative regulators of cell growth and repressors, protein modification and ubiquitin metabolism, more gene transcripts are down-regulated in APOE4/4 AD vs. APOE3/3 AD than in APOE3/4 AD vs. APOE3/3 AD (Supplementary Table 12, 13). Looking at gene expression one obvious difference is the striking number of mitochondrial oxidative phosphorylation and complex I subunits that are down-regulated in APOE3/4 and APOE4/4 AD relative to both APOE3/3 AD and control. Additionally, stress response, chaperone and detoxification, and ER-Golgi transport and cytoskeletal and structural genes are also down-regulated in both APOE3/4 and APOE4/4 AD compared with APOE3/3 AD and APOE3/3 control (Table 11; Supplementary Table 12, 13).
SAGE libraries and data are judged to be of high quality by four criteria. First, there is no GC content bias in all four libraries (Margulies and Innis, 2000). Second, the frequency of both the duplicated ditags and linker contamination is less than 0.5 %, equivalent or below that of published SAGE libraries (Velculescu et al., 1995). Third, comparisons between SAGE libraries demonstrate that the correlation coefficients are in the range of 0.89 to 0.99 (Figure 1) and only 0.41 % to 2.32 % of the tags show changes greater than two-fold (p-value < 0.05) (Table 3). Representative regulated genes were verified by real-time PCR (Table 6) on individual hippocampal RNA samples with the same APOE genotype. Lastly, in separate experiments our SAGE libraries tag count further confirm the qualitative changes in gene expression profiles previously observed in APOE4/4 and APOE3/3 allele-specific AD microarray experiments (Xu et al., 2006).
SAGE data demonstrates, as might be expected, that the hippocampal SAGE patterns in APOE 3/4 AD are similar to those observed in the APOE 4/4 AD. SAGE data indicate that APOE4 gene expression in AD hippocampus induces or significantly up-regulates more than two times and inactivates or down-regulates more than four times the number of genes when compared to APOE3 expression in APOE3/3 control and APOE3/3 AD (Table 3, Fig. 2). EASE molecular functional categorization demonstrates several categories have more elevated genes and all categories have more down-regulated genes in both APOE3/4 AD and APOE4/4 AD vs. APOE3/3 control than in APOE3/3 AD vs. APOE3/3 control (Fig. 3, Table 4). Larger percentages of gene transcripts share up- and down-regulation in hippocampus in both APOE3/4 AD and APOE4/4 AD vs. APOE3/3 control and APOE3/3 AD than in APOE3/3 AD vs. APOE3/3 control (Fig. 5, ,66).
One striking example of significantly up-regulated gene transcripts include tumor suppressors, tumor inducers or tumor associated proteins, and negative regulators of transcription or repressors in both APOE3/4 AD and APOE4/4 AD hippocampus compared with control (Table 7) and APOE3/3 AD (Table 8). Activation and overexpression of these genes (MCC, SMAD2, IGFBP7, DLC1, HIPK2 and MEG3) may increase cell cycle arrest in senescent and/or apoptotic cells, be associated with malfunction of DNA repair and modification (MALAT1, TPM3/TC22, XIST, MUTYH, EMP3, ELAC2 and ECRG4), and inhibit or decrease cell growth and gene transcription (TRIM33/TIF1G, ZHX3, PCOLN3, and DMAP1). Activation of tumor suppressor and oncogenesis genes in APOE3/4 AD and APOE4/4 AD may increase cell cycle arrest and inhibit cell differentiation and cell cycle reentry of neurons in AD brain. Previous studies have found that the relations between tumor suppressors and AD largely involve apoptotic pathways (e.g. p53) (Galderisi et al., 2001). Up-regulation of tumor suppressor related transcription factors is previously reported in incipient AD (Blalock, 2004). Details of gene functions and references in discussion section are contained in Supplementary Discussion online.
A large cluster of gene transcripts involved in synaptic plasticity and synapse formation are significantly down-regulated in both APOE3/4 AD and APOE4/4 AD hippocampus (Table 9; Supplementary table 9-11) (Supplementary Table 12, 13). These results strongly suggest that APOE4 gene expression in the hippocampus of AD patients may inactivate or disrupt gene transcription involved in synaptic plasticity, long term potentiation and depression (EFNB3, EPHA4, SNAP91 and PPP3CA), inhibit or decrease neuronal outgrowth, synaptic regeneration, remodeling and formation of new connections (GAP43, BASP1, NPTX1, NPTXR, SYNGR1, SYNGR3 and SYNPO), reduce synapse stability, and disturb the functions or activities of ion channel and neuronal signaling (PSD95, DLGAP1/GKAP and NMDAR). These data suggest APOE4 expression may likely directly or indirectly inhibit or disrupt neuronal outgrowth, synaptic regeneration, remodeling and formation new connection in the hippocampus of APOE3/4 AD and APOE4/4 AD patients.
SAGE data reveal that multiple active zone proteins involving in presynaptic vesicle docking and fusion are significantly down-regulated in both APOE3/4 AD and APOE4/4 AD compared with APOE3/3 AD (Table 9; Supplementary table 10,11) as well as APOE3/3 control (Supplementary Table 12, 13). Presynaptic terminals are organized into junctions known as active zones where the vesicular protein synaptotagmin (e.g., SYT1) serves as the main Ca2+ sensor for neurotransmitter release (Augustine, 2001;Koh and Bellen, 2003). The highly specialized uptake and exocytosis system of synaptic vesicle traffic has been studied extensively as a model of membrane fusion. The fusion reaction begins with the assembly of syntaxin, SNAP25 and synaptobrevin (Singh et al., 2004) into a core complex that serves as a SNAP receptor, or SNARE. These down-regulated gene transcripts play the critical roles in regulating synaptic membrane exocytosis (RIMS2 and RIMS3), serving as calcium sensors (SYT1 and SYT13), forming both SNARE complex (VAMP2 and NAPB) and synaptic vesicle (SV2A and VAPB), and scaffolding presynaptic active zone assembly or functions (RIMBP, DNM1, CPLX1 and PPFIA4/Liprin) (Augustine, 2001;Koh and Bellen, 2003).
These losses of genes encoding proteins functionally related to synaptic vesicle docking and fusing and synaptic plasticity appear to be APOE4 allele(s) specific and are found only in APOE3/4 AD and APOE4/4 AD (Table 9). In AD loss of synapses correlates with cognitive decline (DeKosky and Scheff, 1990;Terry et al., 1991). Reduction of synaptic vesicle trafficking related genes including synaptophysin, synapsin, synaptotagmin, AP180, DNM1 and RAB3A have been reported previously in AD (Eikelboom et al., 2000;Ho et al., 2001;Bouhouche et al., 1999;Yao et al., 2003). A loss of afferents from entorhinal cortex are believed to underlie the synapse loss seen in early AD (Scheff et al., 2005). Coleman and Yao (Coleman and Yao, 2003) provide evidence that in addition to loss of synapses, AD synapses may not function effectively or appropriately by decreasing transcripts related to synaptic vesicle trafficking (Yao et al., 2003). These findings suggest that losses of multiple presynaptic active zone proteins in AD with APOE4 allele(s) may directly or indirectly disrupt normal synaptic vesicle docking and fusing and lead to disturbance of neuronal circuitry and activities and consequently result in cognitive impairment.
Functional clusters of gene transcripts involved in axonal guidance, neurite outgrowth and neurogenesis are significantly down-regulated (or inhibited) in both APOE3/4 AD and APOE4/4 AD hippocampus compared with APOE3/3 control (Table 9) and AD hippocampus (Supplementary Table 12, 13). These downregulated gene transcripts are involved in regulating neuronal axon guidance, dendritic spine morphogenesis and growth cones (SLIT1, DOK-6, NGEF, PPFIA4 and CDH10); modulating neurite outgrowth, neuronal polarity and neuronal development (THY1, GSK3B, GAS7, PPP1R9B, AF1q, TM4SF2, RUSC1 and DULLARD); and mediating neurogenesis, axonal regeneration and plasticity and neurodegeneration (RTN4R, ELAVL3, ITM2B, ITM2C and TPP1). These findings are significant in that they may provide some insight into previous observations that APOE4 protein decreases neurite outgrowth and neuronal branching compared with APOE3 (Nathan et al., 1994).
A large cluster of gene transcripts modulating calcium signaling homeostasis and encoding both excitatory (AMPA and NMDA) and inhibitory (GABA-A and GABA-B) neurotransmitter receptors are significantly down-regulated in APOE3/4 AD and APOE4/4 AD compared with APOE3/3 control (Table 10, ,11)11) and APOE3/3 AD (Supplementary Table 12, 13). The gene transcription of CAMK2A, CAMK2D, CALM2 and CALM3 are also down-regulated in APOE4/4 ADs compared with APOE3/3 ADs in microarray analysis (Xu et al., 2006). Calcium signaling is crucial for several aspects of plasticity at glutamatergic synapses and in the postsynaptic density (PSD). Intracellular calcium is one of the most extensively disrupted transduction systems in AD (Prolla and Mattson, 2001;Stutzmann, 2005). Excess glutamate causes neurotoxicity and may contribute to cellular damage in neurological disorders such as stroke, trauma (Harada and Sugimoto, 1998;Su et al., 2003), Alzheimer disease (AD), amyotrophic lateral sclerosis (ALS) (Rothstein et al., 1995), and Huntington disease (Choi, 1988;Kanai et al., 1993). The loss and/or down-regulation of calcium signaling and both excitatory and inhibitory neurotransmitter receptors in the hippocampus of AD patients carrying APOE4 allele(s) could potentially affect or decrease LTP, learning and memory as well as increase neuroexcitotoxicity, apoptosis and neurodegeneration.
The functional cluster of signal transduction pathways contains the most down-regulated gene transcripts in APOE3/4 and APOE4/4 AD compared with APOE3/3 AD (Supplementary Table 12, 13) and APOE3/3 control (Table 10). These signaling molecules regulate a variety of signal transduction pathways and cascades including mitogen-activated protein (MAP) kinase cascades (MAPK9, MAPK8IP2 and MAP2K1), NF-κB signaling cascades (BTCR, T1/NIBP, COMMD7), protein kinases (PRKACB, PRKAR1B, CSNK2B, ILK, NLK, ARAF1), protein tyrosine phosphatases (PTPN5, PPFIA4 and PTPRN), and G-protein signaling (GNG3, GNG, GNB5, GNAO, GNAZ, GPR51/GABBR2, RGS4 and RGS6). Several proteins may mediate regulatory signal transduction pathways in neurons to modify neuronal cytoarchitecture (TAGLN3/hNP22) and promote actin repolymerization and/or cytoskeletal reorganization (CHN1, CDC42BPA and CAP).
Several gene transcripts encoding proteins that protect cells and neurons from heat shock and oxidative stresses and detoxification are downregulated in hippocampus of APOE3/4 AD and APOE4/4 AD compared with APOE3/3 AD and control (Table 11, Supplementary Table 12, 13). These proteins function as chaperones in the cellular heat shock response to remove denatured or incorrectly fold proteins (NAJA4, DNAJC8, STCH, HSPCB and AHSA2), detoxify hydrogen peroxide to protect against oxidative damage or stress (OXR1 and GPX1), suppress toxicity by sequestering poly (Q) and/or its aggregation (MLF2 and ASNA1), and promote clearance of apoptotic cells (C1QA). It is essential for cells to control the production and activity of such molecules as reactive oxygen species (ROS) in order to protect the genome and regulate cellular processes such as stress response and apoptosis (Elliott and Volkert, 2004).
In AD pathogenesis and other neurodegenerative disorders, increasing evidence implicates mitochondrial dysfunction and defects in oxidative phosphorylation (OXPHOS), leading to oxidative damage and defects in energy metabolism manifested by down-regulation in mitochondrial number (Ojaimi and Byrne, 2001;Manczak et al., 2004). A striking number of mitochondrial oxidative phosphorylation and complex I subunits (e.g. cytochrome c oxidases, NADH dehydrogenases, ATP synthases, etc.) are significantly down-regulated in APOE3/4 AD and APOE4/4 AD patients compared with APOE3/3 AD (Supplementary Table 12, 13) and APOE3/3 controls (Table 12). These SAGE data are consistent with recent published microarray results (Xu et al., 2006). COX defects have been suggested as a primary etiologic event (Davis and Weeks, 1997) and this study suggests that in APOE3/4 AD and APOE4/4 AD compared to APOE3/3 AD the greater global defects of mitochondrial energy metabolism and mitochondrial/oxidative damage may contribute to increased risk.
In summary, SAGE data, taken in conjunction with recent results on APOE allele-specific expression from multiple patients using microarrays (Xu et al., 2006) together, suggest that gene expression in the hippocampus of AD APOE3/4 and 4/4 patients are different from that observed in APOE 3/3 AD patients and offers support for the hypothesis that APOE3/4 and APOE4/4 gene expression are related and may regulate similar gene pools with associated functions compared to the APOE3 gene expression.
Human brain tissues were collected in the Kathleen Price Bryan Brain Bank, in the Duke University Alzheimer Disease Research Center, and the Center for Human Genetics' Brain Bank, Duke University Medical Center, following the rapid autopsy protocol (Hulette et al., 1997). The hippocampus was dissected at the time of autopsy and matching 100-200 mg portions of CA 1-4 was removed and used for RNA isolation and expression studies. We chose AD patients of apolipoprotein E4/4 (APOE4/4), APOE3/4, and APOE3/3 genotypes and controls with the APOE 3/3 genotypes judged cognitively and pathologically normal. Clinical information on all patients was collected from clinic and hospital records. The pathological diagnosis of AD was established according to CERAD criteria (Mirra et al., 1991) and the degree of AD pathological changes was staged according to Braak (Braak and Braak, 1991). We analyzed selected AD patients with pathological changes at the Braak and Braak stage IV or V (B&B stage IV or V) and compared them with cognitively and pathologically normal controls (B&B stage I) (Table 1). We chose hippocampal samples with mostly short post-mortem delay. RNA was analyzed using gel electrophoresis, bioanalyzer analysis, and judged by microarray analysis to have excellent 5′ to 3′ ratio.
Total RNA was isolated from frozen hippocampal samples of AD patients and controls using TRIzol reagent (Invitrogen, Carlsbad, CA) as previously described (Xu et al., 2006). For SAGE library construction, we used standard protocols as described by Velculescu et al (1995) with minor modifications. Briefly, SAGE was performed with 10 μg total RNA isolated from human brain hippocampus samples as outlined above. The cDNA was prepared using the SuperscriptII cDNA synthesis kit (Invitrogen) with gel-purified 5′-biotinylated Oligo(dT)18 (Integrated DNA Technologies, Coralville, IA), according to the manufacturer's protocol. NlaIII and BsmFI restriction enzymes (New England Biolab, Beverly, MA) were used for tag generation. BsmFI digestion was performed at 37°C for 2.5 h (instead of 65 °C) using 40 units BsmFI in a 300 μl reaction volume with supplied buffer. After a 3-hour concatemerization step, concatemers were heated at 65 °C for 10 minutes, following by 2 minutes on ice to enhance cloning efficiency. Purified concatemers were subsequently cloned in the SphI site of pZero-1 (Invitrogen) and transformed in competent ElectroMax DH10B cells (Invitrogen) using a 0.1 cm cuvette and the Gene Pulser II (BioRad). Individual SAGE library clones were selected and PCR amplified using 96-well format Qiagen Real minipreps, and sequenced with ABI 3700 capillary sequencer using BigDye chemistry.
SAGE tags (10 bp) were extracted from the PHD files with eSAGE software using a threshold value of PHRED 20 for each base (Margulies and Innis, 2000). The SAGE tags were extracted from individual sequence files from each SAGE library (Table 2). The SAGE tag databases were compared with each other, such as AD vs. control and APOE4/4 AD vs. APOE3/3 AD SAGE tag databases, etc. using eSAGE software (Table 3). The compared SAGE databases were mapped to UniGene Homo Sapiens tag-to-gene mapping set (build 182, www.ncbi.nlm.nih.gov/UniGene).
The Chi-square test and Fisher exact test were employed to test the difference in tag counts between two samples as previously described (Man et al., 2000). For each particular Tag A, we arranged the data in a 2×2 table.
|Library 1||Library 2|
The tag counts for Tag A in sample 1 and sample 2 are designated as n11 and n12, respectively. The n21 and n22 are the sum of the remaining tags in library 1 and library 2, respectively. The Chi-square statistic (χ2) for the 2×2 table is computed by the following formula and compared to 1 degree of freedom at the 0.05 type I error rate.
In general, Chi-square analysis requires an expected cell count ≥5. To be conservative, when any of the expected count values in the 2×2 table (n11, n12, n21, n22) are less than or equal to 10, we performed the Fisher exact test in addition to the Chi-square test. In the Fisher exact test, we fixed the marginal sum (i.e. N1., N2., N.1, N.2) and created all possible tables with the same marginal sum. The probability of observing each possible table configuration is computed as below.
The two-sided p-value of the Fisher exact test is the sum of all probabilities that are less than or equal to the probability derived from the original observed table. We used these methods to compare SAGE tag abundance for each pair of AD and control libraries.
Since p-values were calculated for several thousands of genes simultaneously, we applied Benjamini and Hochberg's (Benjamini and Hochberg, 1995) false discovery rate (FDR) procedure to correct for multiple testing. We simultaneously test n null hypotheses H1, H2, …, Hn on the basis of independent chi-square or Fisher exact tests. The p-values were ordered as P(1) ≤ P(2) ≤ …≤ P(n). If k is the largest integer such that , then we reject all Hi for i<k, where po = proportion of true Hi. For this analysis, we assume α = 0.05 and p0=1 (i.e. all Hi are true). We have used two different values for the number of independent tests.
We assigned differentially expressed gene transcripts to functional categories by two methods. One method assigned transcripts to broad functional categories defined on the basis of gene function derived from literature searches. The other method used the software tool, the Expression Analysis Systematic Explorer (EASE; http://david.abcc.ncifcrf.gov), to assign identified gene transcripts to “GO: molecular function” categories of the GeneOntology Consortium (www.geneontology.org). We used the GO system to quantify over- and under-represented (up- and down-regulation) transcripts relative to total genes within a functional category and compare between different SAGE libraries. Since the GO system does not provide a functional designation for all genes, our functional interpretations relied primarily on the extensive literature search in PubMed and sequence search in NCBI and other HGP databases.
We confirmed selected SAGE data in individual APOE allele-specific ADs and controls by quantitative real-time PCR (qRT-PCR). Hippocampal total RNA (10 ug) was converted to cDNA using the High Capacity cDNA Archive Kit (Applied Biosystems, Foster City, CA) according to the manufacturer's protocol. Assays-on Demand™ Gene Expression primer sets for qRT-PCR were purchased from Applied Biosystems (Table 6). The primer sequences can be obtained from the authors upon request. Four housekeeping genes (GAPD, GUSB, HPRT1, APLPO) were used as endogenous controls (or references). Real-time PCR was performed using TaqMan Universal PCR Mater Mix according to TaqMan Gene Expression Assay protocol (ABI). Each reaction was performed in quadruplicates, and non-RT (without reverse transcriptase) and non-RNA template controls were included. The relative amounts of specifically amplified cDNAs in APOE allele-specific ADs compared with APOE3/3 controls were calculated using comparative CT method (ΔΔCT method) using Arithmetic Formula: 2 −ΔΔCT (User Bulletin #2, ABI). The average CT value for each genotype of AD cases was calculated from individual samples in each APOE genotype group (Table 1) on four independent experimental replicates. The mean ΔCT value was calculated by formula: ΔCT = CT target - CT reference. The ΔΔCT of target gene was calculated by formula: ΔΔCT = ΔCT target - ΔCT calibrator. We use average GAPDH CT value as CT reference and average ΔCT value of APOE3/3 control as ΔCT calibrator.
We thank the Alzheimer disease patients and their families. We also thank the clinical and research personnel of the Center for Human Genetics at Duke University Medical Center and the Joseph and the Kathleen Bryan ADRC. This work was supported by the 2001 Louis D. award from the Institut de France; grants NS31 153, AGO5 128, AG13308, AG1 1268, AGIO123, AG19085, MH52453, MH59528, NSO31 153, NS26630, and NS3 9764 from the National Institutes of Health.
The AD SAGE tag count and lists were submitted to NCBI Gene Expression Omnibus (GEO). The GEO accession number is GSE6677.
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final citable form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.