Jianping Jia, Department of Neurology, Xuan Wu Hospital of the Capital Medical University, Beijing 100053, China. Tel.: 0086 10 8319 8730; fax: 0086 10 83171070; e-mail: firstname.lastname@example.org; email@example.com
We previously described an association between Alzheimer’s disease (AD) and a single-nucleotide polymorphism −980C/G (rs3754048) in the promoter of the anterior pharynx-defective-1a (APH-1A) gene. Here, we examine the potential of this −980C/G polymorphism to affect APH-1A transcription and confer a risk of AD. We validated the presence of APH-1A promoter polymorphism −980C/G in other two Chinese cohort sets (450 AD and 450 controls). Subsequently, we measured APH-1A mRNA and protein levels and γ-secretase activity in C or G allele carriers. Finally, we examined the polymorphism’s transcriptional function using a dual-luciferase reporter assay and also tracked transcription factor binding to the variant promoter sequence with electrophoretic mobility shift assays (EMSAs). We found that the APH-1A levels and γ-secretase activity were higher in individuals carrying allele G. The G allele increased APH-1A transcriptional activity significantly in both N2A cells and HEK293 cells. The EMSA revealed an increased binding of the transcription factor Yin Yang 1 (YY1) to allele G. Overexpression of YY1 resulted in an activation of the APH-1A promoter (2.7-fold). Specific YY1 siRNA led to decreases in APH-1A promoter activity and mRNA and protein levels. Our data indicate that the APH-1A promoter polymorphism −980C/G might alter the binding ability of YY1 transcription factor, resulting in an increased level of APH-1A and γ-secretase activity. These factors further facilitated β-amyloid (Aβ) 42 generation and ultimately modified patients’ susceptibility to AD. The involvement of transcription factor YY1 might be a novel mechanism for the development of AD.
Alzheimer’s disease (AD), the most prevalent neurodegenerative disease of humans, now afflicts about 20 million people worldwide (Selkoe & Schenk, 2003). The progressive loss of memory and declining cognitive function caused by AD ultimately lead to decreased physical function and death.
Deposition of β-amyloid (Aβ) 42 in the brain is a hallmark of AD (Patel et al., 2005). Aβ42, the major component of neuritic plaques, is generated from amyloid precursor protein (Blandini et al., 2006) via sequential cleavages by β-secretase and γ-secretase complex (Zheng & Koo, 2006). The γ-secretase complex consists of at least four components: presenilin, nicastrin, anterior pharynx-defective-1 (APH-1), and presenilin enhancer-2 (Kimberly et al., 2003). Goutte et al., (2002) first identified APH-1 as a gene in the early C. elegans embryo. APH-1, which forms a stable subcomplex with nicastrin and contributes to the stabilizing and trafficking of the γ-secretase complex, is a key factor in γ-secretase activity. Two homologs of APH-1 were identified in human, i.e., APH-1A and APH-1B (Saito & Araki, 2005). Down-regulation of APH-1A and APH-1B by small RNA interference alters the formation of multimeric complex and greatly reduces the production of Aβ (Luo et al., 2003). As APH-1A is the principal mammalian APH-1 isoform within γ-secretase complexes (Ma et al., 2005), the overexpression of APH-1A can increase γ-secretase activity and cellular Aβ content (Kim et al., 2003).
Accumulating evidence suggests that polymorphic variability within a gene promoter could affect gene expression and was associated with neurodegenerative disorders (Cong & Jia, 2011; Wang & Jia, 2010). We therefore hypothesized that alterations in the APH-1A promoter region altered APH-1 expression and affected the activity of γ-secretase, which consequently influenced Aβ production. Our group previously performed a thorough search for sequence variations in the promoter of APH-1A gene in samples from non-AD controls and patients with AD (Wang & Jia, 2009). We detected two new single-nucleotide polymorphisms (SNPs), but only the polymorphism −980C/G (rs3754048) in APH-1A showed a significant association with AD (We number nucleotides based on the translation start site expressed as +1). For further validation, we investigated the APH-1A promoter polymorphism −980C/G in two additional separate cohorts of patients and controls to identify the biological basis for this association.
APH-1A promoter polymorphism −980C/G
APH-1A allele and genotype distributions were in Hardy–Weinberg equilibrium. Notably, the patients with AD from XW Hospital exhibited a significantly greater frequency of the APH-1A−980C/G allele and genotype than controls at that hospital (Table 1). Additionally, in logistic regression analysis, the patients’ risk of AD was associated with G allele to a significant extent after adjusting for age, gender, education, and APOEε4 status (OR = 3.21, 95% CI = 1.81–5.69). Samples from the GMSA Hospital, when genotyped for APH-1A−980C/G, displayed a risk of AD in the GG group that was 3.66-fold higher than the CC+CG genotype (OR = 3.66, 95% CI = 1.88–7.14). Meta-analysis for −980C/G with two hospital sets revealed a significant association with AD (allele P =0.006, OR = 1.61, 95% CI = 1.33–1.94; genotype P = 0.0001, OR = 2.65, 95% CI = 1.85–3.80; XW Hospital, Weight% = 58.44%, GMSA Hospital, Weight% = 41.56%). Results as shown in Table 2 demonstrated that these differences of −980C/G genotype frequencies between AD and controls remain significant in APOEε4 (+) subgroups. After adjusting gender, age, and education, a much higher OR for the risk of AD was seen when comparing GG with CC+CG genotype in APOEε4 carriers (Table 2).
Table 1. Distribution of the polymorphism in patients with AD and controls
Nominal P value
GG vs. CC+CG
AD, Alzheimer’s disease; CI, confidence interval; MAF, minor allelic frequency; OR, odds ratio; GMSAH, the Affiliated Hospital of Gui Yang Medical School; XWH, Xuan Wu Hospital. P <0.05 is indicated in bold.
†Data were calculated by logistic regression, adjusting for gender, age, education, and APOEε4 status.
Table 2. AD risk analysis of rs3754048 in APOEε4 (+) subjects
In subjects with APOEε4 +
P value(GG vs. CC+CG)
AD, Alzheimer’s disease; CI, confidence interval; OR, odds ratio; GMSAH, the Affiliated Hospital of Gui Yang Medical School; XWH, Xuan Wu Hospital.
†Data were calculated by multinomial logistic regression, adjusting for gender, age, and education.
The G allele enhances APH-1A mRNA and protein expressions
To determine whether −980C/G polymorphism was associated with the expression level of APH-1A, we used real-time PCR to quantify the APH-1A mRNA in fresh lymphocytes isolated from peripheral blood of 167 individuals from XW Hospital. The mRNA expression of APH-1A was increased by 13% in 65 patients compared with 102 controls, but this difference did not reach the level of significance (P = 0.08; Fig. 1A). However, the differences were significant when −980C/C and −980G/G (P = 8.01E-08) or −980G/G and −980C/G (P = 0.002) were compared (Fig. 1B).
ELISA was used to measure the relative protein levels of APH-1A in fresh lymphocytes isolated from peripheral blood of 70 controls and 65 patients with AD. Similar to mRNA results, remarkable higher protein levels of APH-1A were also detected in individuals with −980G/G than those with −980C/C or −980C/G (Fig. 2). These suggested G allele plays an important role in such a protein overexpression of APH-1A. This was supported by the fact that APH-1A levels corresponded well to the numbers of G alleles.
The G allele heightens γ-secretase activity
Because APH-1A is crucial for γ-secretase activity, which in turn may influence the generation of Aβ, we examined the γ-secretase activity in lymphocytes from subjects with different APH-1A−980 genotypes. We found that γ-secretase activity was, in fact, associated with −980C/G genotype, i.e., those with −980G/G showed higher γ-secretase activity (median = 41997.13 IU mL−1) than the group with −980C/C (median = 39602.75 IU mL−1) (P = 0.001) or with −980C/G (median = 40607.65 IU mL−1) (P = 0.006; Fig. 3).
The−980C/G polymorphism impacts transcriptional activity of human APH-1A promoter
As the −980C/G polymorphism could alter APH-1A transcriptional activity, we evaluated this possibility by generating two luciferase reporter vectors (pGL3 basic) spanning −1226 to +395bp within the APH-1A promoter region (Fig. 4A). These vectors had identical sequences except for the −980C/G (Fig. 4B). The outcome of transfecting these vectors into mouse neuroblastoma Neuro-2A (N2A) or human embryonic kidney293 (HEK293) cells was that the −980 GG-containing promoter drove luciferase expression in a statistically significantly greater manner than the −980 CC-containing promoter (Fig. 4C; in N2A cells, P = 0.001; and in HEK293 cells, P = 0.010).
Transcription factor binding is allele specific
Whether the genetic variants of −980 would alter DNA–protein interaction was determined by using electrophoretic mobility shift assays (EMSAs). Labeled double-stranded DNA probes, one containing the wild-type sequence (C/C) (C probes) and the other a risk-type sequence (G/G) (G probes), were used for the formation of DNA–protein complexes with nuclear extracts from rat primary neurons, N2A or HEK293 cells (Fig. 5A). The G probes generated a stronger binding complex band with nuclear extract than the C probes. Also, competition with a 100-fold excess of the unlabeled C probes did not completely inhibit binding of the complex to the −980G allele (Fig. 5B).
The −980C/G transversion appeared to alter the consensus sequences of several transcription factors on the basis of analyses with Genomatix software. To identify which transcriptional factor binds to this susceptible allele, we added excess amounts of unlabeled oligonucleotides corresponding to consensus sequences of various transcriptional factors (AP1, AP4, USF-1, YY1, or HIF-1) as competitors. The specific band was abolished by the addition of 50-fold excess AP1, AP4, USF-1, or YY1 consensus sequence (Fig. 5C). To confirm which transcription factor was involved in forming the specific band, we performed supershift experiments by adding AP1, AP4, USF-1, or YY1 antibodies to the binding reactions (Fig. 5D). The addition of YY1 antibody weakened the G allele-specific band and supershifted a DNA–protein complex band. Similar results were obtained using nuclear extracts from N2A cells or HEK293 cells (data not shown).
Transcription factor YY1 upgrades APH-1A promoter activity and expression
When cotransfected with YY1 expression plasmids, the APH-1A promoter yielded reporter activity that increased 2.03-fold (P = 1.13E-08) in N2A. The presence of reporter plasmid containing the risk-type sequence (G) had an even more prominent effect (Fig. 6). Similar result was obtained in HEK293 cells transfected with these plasmids (Fig. 6). The detailed data are described in the Table S2.
To test whether the YY1 transcription factor was indeed responsible for these heightened levels of APH-1A, endogenous YY1 was inhibited with specific siRNA. As predicted, luciferase activity driven by plasmids carrying the APH-1A promoter was significantly attenuated by the introduction of YY1 siRNA (P =1.01E-07) (Fig. 6). The reduction of YY1 levels dramatically decreased APH-1A mRNA and protein expressions (mRNA levels, P =0.032; protein levels, P =0.009; Fig. 7).
In the present study, we have confirmed our previous conclusion that −980C/G polymorphism in the promoter of APH-1A gene is associated with AD (Wang & Jia, 2009) by extending our investigation to a much larger group of 450 patients with confirmed AD. Here, we not only reassert the genetic association of this −980C/G polymorphism in two separate patient populations but also strengthen our previous finding by showing that the binding of YY1 transcription factor raises the mRNA and protein levels of APH-1A and subsequently heightens γ-secretase activity. It may promote Aβ generation and contribute to the risk of AD. Until now, no other reports have substantiated the role of the APH-1A promoter polymorphisms in the pathogenesis of AD.
Dysfunctions in genetic and molecular mechanisms in the periphery have been confirmed as participants in the systemic perturbations of patients with AD (Ray et al., 2007). Ray et al. (2007) suggested that differential gene expression patterns in blood cells can predict probable AD. The brain-specific glutathione S-transferase mu 3 (GSTM3) protein is unique in that it accumulates in senile plaques and neurofibrillary tangles (Tchaikovskaya et al., 2005). Diminished transcript levels of GSTM3 were reported in the AD-affected hippocampus (Blalock et al., 2004) and blood mononuclear cells (Maes et al., 2010). Johnston et al. (2008) revealed that β-secretase activity was elevated in postmortem brain tissue as well as in peripheral blood platelets of patients with AD. They suggested that the increased β-secretase activity observed in platelets reflects a systemic increase in activity. PI3K/Akt activation has been detected in brains and lymphocytes from patients with AD (Munoz et al., 2008). Oxidative damage to DNA is an early event in the pathogenesis of AD and has been documented within lymphocytes and affected brains of patients with AD (Blandini et al., 2006). Expression profiling of peripheral blood cells thus provides a biological window for investigating complex etiopathogenetic mechanisms in the central nervous system (Munoz et al., 2008). Using this strategy, we found that there were significantly higher mRNA and protein levels of APH-1A in the lymphocytes of individuals carrying the APH-1A−980 risk allele G. Higher levels of APH-1A may contribute to an increased risk of AD.
APH-1A is required for the efficient binding of substrates, which is important for γ-secretase activity (Hu & Fortini, 2003; Chen et al., 2010). Interest in γ-secretase comes from the fact that this multiprotein complex is responsible for the cleavage of APP that generates the Aβ, one of the primary components of amyloid plaques in AD (De Strooper et al., 2010). Moreover, an alteration in γ-secretase activity is involved in the pathogenesis of AD (Placanica et al., 2009). Studies have suggested that APH-1A is a limiting factor for γ-secretase activity and that an increased APH-1A concentration alone may be sufficient to elevate the γ-secretase activity (Wang et al., 2006). Consistent with these observations, we found that the APH-1A−980 G/G genotype was associated with elevated γ-secretase activity. This enhancement suggests that APH-1A−980C/G promoter polymorphism functioned in modulating APH-1A expression and γ-secretase activity, which could facilitate Aβ generation and eventually increase susceptibility to AD.
Next, we analyzed the molecular mechanisms underlying the role of APH-1A−980C/G polymorphism in the increased risk of AD. The effect of the C- and G alleles of the −980 polymorphism on transcriptional activity was tested by instituting a dual-luciferase reporter assay. We found that the risk allele G displayed a strikingly higher activity than the wild-type allele C in both the neural and nonneural cell lines. Presumably, then, the G allele possessed higher transcriptional activity and had overexpressed APH-1A, which might have played a pivotal role in the development of AD.
It is important to clarify why the −980C/G polymorphism resulted in higher APH-1A transcriptional activity. Binding sites of transcription factor HIF-1 and AP4 in the APH-1A promoter have been identified as modifiers of APH-1A expressions (Wang et al., 2006). Using EMSAs, we found that the −980C/G polymorphism affected APH-1A transcriptional activity because of altered YY1 binding affinity. YY1 is a widely expressed C2H2 zinc finger transcription factor that can activate or repress a number of gene promoters and regulate gene expression (Jiang et al., 2008). A higher binding affinity of YY1 to the G allele was identified in this study, possibly causing the activation of APH-1A promoter activity in patients with AD who bear the −980G/G genotype. We also found that the specific siRNA of YY1 significantly decreased APH-1A promoter activity and down-regulated APH-1A mRNA and protein expressions. This outcome supports our hypothesis that YY1 is an important regulator of APH-1A promoter activity and its expression. YY1 was previously reported to be involved in the regulation of several important AD-related genes. Briefly, (Nowak et al. (2006) demonstrated that YY1 acted as an activator of the BACE1 (β-site of APP-cleaving enzyme 1, BACE1) promoter. BACE1 expression and enzymatic activity are increased in AD-afflicted brains. YY1 also can bind to the FE65 minimal promoter and increase its transcription (Zambrano et al., 1997), thereby inducing a robust Fe65-dependent increase in Aβ generation (Sabo et al., 1999). From our results, we suggest that the −980C/G polymorphism enhanced APH-1A expression by increasing its affinity for binding to YY1 and, as a consequence, promoted AD. Therefore, the potential for the use of YY1 inhibitors might be an AD prevention strategy, which warrants further exploration and development. It will also be interesting in future studies to reveal the mechanism for controlling APH-1A-mediated γ-secretase activity by YY1 inhibitors. It may be important for providing useful information for the prevention and/or treatment of AD.
There are limitations present within our study. First, although we found that there were significantly higher APH-1A levels in individuals carrying the APH-1A −980 risk allele G, further studies in larger populations are certainly needed to confirm these results. Second, AD has a complex pattern of inheritance, and multiple genes may contribute to the disease susceptibility. In addition to such genetic risk factors, various environmental factors have been proposed to alter the risk of developing AD as well as to affect the rate of cognitive decline in patients with AD. These highlight the need to explore the effect of gene/gene or gene/environment interaction related to AD in our future genetic studies.
In conclusion, by extending our previous genetic association study, we have successfully identified the risk-conferring polymorphism in the APH-1A gene promoter. A primary role of the polymorphism was to alter YY1 binding affinity, resulting in increased mRNA and protein expressions of APH-1A and heightened γ-secretase activity among individuals carrying allele G. As γ-secretase plays an important role in the generation of Aβ, our findings suggest that the functional SNP of APH-1A may influence γ-secretase activity and place patients at risk of AD. The modulation of APH-1A expression via the YY1 transcription factor offers a way to illuminate the role of this γ-secretase component in normal biology and could provide a novel strategy for the target in developing new drug for AD.
This study included two sets of samples from patients with AD and from healthy controls, collected independently and consecutively from Xuan Wu (XW) Hospital of Capital Medical University in Beijing (250 patients/250 controls) and the Affiliated Hospital of Guiyang Medical School (GMSA Hospital, 200 patients/200 controls), between the years 2007 and 2010. For assays of gene expression and γ-secretase activity, we obtained 6 mL of fresh blood samples from 102 healthy volunteers (50 men and 52 women with a mean age of 71.22 ± 11.21 years) and 65 patients (32 men and 33 women with a mean age of 72.13 ± 12.41 years) from XW Hospital. For assays of protein expression, we obtained 12 mL of fresh blood samples from 70 healthy volunteers (38 men and 32 women with a mean age of 69.92 ± 10.17 years) and 65 patients (30 men and 35 women with a mean age of 70.31 ± 10.21 years) from XW Hospital. The detailed information about subjects is described in the Data S1.
Genomic DNA was isolated from peripheral blood samples by a salting-out procedure, and all DNA samples were normalized to 50 ng μL−1. APH-1A (NM_001077628.1) promoter polymorphism (rs3754048) genotyping was performed by direct sequencing. APOE genotyping adhered to published protocols (Zuo & Jia, 2009).
Lymphocyte isolation and culture
Peripheral blood collected from each subject was heparinized and used as the source of lymphocytes, which were isolated immediately as described (Zhou & Jia, 2010) to avoid freezing and long-term storage. After washing, cells were suspended in RPMI-1640 medium, supplemented with heat-inactivated FBS (10%, v/v), l-glutamine (2 mm), penicillin (100 μg mL−1), and streptomycin (100 μg mL−1), then maintained in a humidified 5% CO2 incubator at 37 °C. Cell viability was always >90%.
Real-time quantitative PCR and allele-specific expression assay
Total RNA was isolated from peripheral blood lymphocytes of 167 individuals and then converted to cDNA using Superscript II (Invitrogen, Carlsbad, CA, USA). Total APH-1A mRNA was measured by real-time PCR. Reactions were performed in 25-μL volume units that included diluted cDNA samples, primers, and SYBR Green I Mastermix (Applied Biosystems, Carlsbad, CA, USA). Expression of APH-1A was normalized relative to β-actin levels (denoted as ΔCt). Each sample of β-actin and APH-1A was tested four times. Results were analyzed by the ΔΔCt method.
ELISA and protein expression assay
An indirect ELISA method was carried out to quantify relative protein levels of APH-1A in extracts from lymphocytes. Each well of the ELISA plate was coated with 4-μg protein sample for 2 h at room temperature. After washing with phosphate-buffered saline (PBS), coated wells were blocked and then incubated with primary antibodies (mouse anti-APH-1A 1:300; mouse anti-β actin 1:500; Santa Cruz Biotechnology, Santa Cruz, CA, USA) overnight at 4 °C. Color was developed by tetramethylbenzidine (Beyotime, Shanghai, China) used as a substrate, and reaction was stopped by the addition of 100 μL of H2SO4. Optical density (OD) was measured at 450 nm using Paradigm (Beckman Coulter, Brea, CA, USA). The protein level was expressed as the ratio of the OD of protein of interest to actin. Measures were made in duplicate, and interassay variability was <9%.
Analysis of γ-secretase activity in peripheral blood lymphocytes
γ-secretase activity was measured using a γ-secretase activity kit (R&D Systems, Minneapolis, MN, USA). Thirty micrograms of lymphocytes protein was incubated in reaction buffer with 5-μL substrate in the presence of DMSO for 2 h at 37 °C in the dark. γ-secretase cleavage of the substrate was detected by excitation of the fluorophore at 340 nm, and emissions were monitored at 500 nm using Paradigm (Beckman Coulter). Activity was expressed as fluorescent intensity at 500 nm.
Luciferase reporter assays
The APH-1A promoter-luciferase reporter plasmids containing either the wild-type sequence C or the risk sequence G were constructed. The APH-1A expression plasmids were purchased from Proteintech Group Inc. (Chicago, IL, USA). These plasmids were confirmed by DNA sequencing.
Cortical neurons were harvested from rat pups (postnatal day 0) and maintained in neurobasal media with supplement of B27, 100 g mL−1 penicillin, 100 g mL−1 streptomycin (Invitrogen), and 0.5 mm glutamine. Mouse neuroblastoma Neuro-2A (N2A) or human embryonic kidney 293 (HEK293) cells were propagated in RPMI 1640 medium with 10% fetal bovine serum, 100 g mL−1 penicillin, and 100 g mL−1 streptomycin.
For transient transfection, cells were seeded at 1.5 × 104 cells per well in a 96-well plate 24 h before transfection. Cells were cotransfected with 150-ng reporter plasmids and 3-ng pRL-TK constructs using Lipofectamine 2000 (Invitrogen). The pRL-TK plasmid (Promega, Madison, WI, USA) was used as internal reference. Empty pGL3-Basic vector was used as a negative control and pGL3-Control vector as a positive control. Transfected cells were cultured for 48 h, washed with 200 μL PBS, and lysed with 20-μL passive lysis buffer (Promega). Luciferase activities of firefly (LAF) and Renilla (LAR) were measured sequentially using a dual-luciferase reporter assay system (Promega) and a model GloMax™ 96 Microplate Luminometer (Promega). To correct for transfectional efficiency, DNA uptake, and expression efficiency, the relative luciferase activity (RLA) was calculated as RLA = LAF/LAR.
Electrophoretic mobility shift assays
Nuclear proteins were extracted by using a nuclear extraction kit (Pierce, Rockford, IL, USA). EMSAs were performed using gel shift assay systems (Promega) under the guidelines provided. For each allele of −980G/C, pairs of single-stranded, biotinylated, and unlabeled 26-bp oligonucleotides (Shanghai Sangon of China, Shanghai, China) were allowed to anneal to generate double-stranded probes. Twenty moles of labeled probes was incubated for 20 min with 10 μg of nuclear extract prepared from the primary neurons, N2A and HEK293, in a freshly made binding buffer supplemented with poly(dI-dC) and protease inhibitors. Competition experiments were performed using a 50-fold molar excess of unlabeled probe. The biotinylated fragments were detected with a LightShift Chemiluminescent EMSA kit (Pierce) according to the manufacturer’s instructions.
For competition experiments, several transcription factor consensus oligonucleotides were obtained from Santa Cruz Biotechnology. A rabbit polyclonal antibody transcription factor was used for supershift in the EMSA. Additionally, transcription factors were analyzed with Genomatix software (Genomatix Software GmbH, Munich, Germany, http://www.genomatix.de/products/index.html).
Gene silencing with siRNA
The transcription factor Yin Yang 1 (YY1) siRNA lentiviral particles were purchased from Santa Cruz Biotechnology. Cells were seeded in DMEM medium with 10% fetal bovine serum without antibiotics. The next day, cells were incubated with 4 μg mL−1 polybrene containing 1 × 105 infectious units of virus according to the manufacturer’s instructions (Santa Cruz Biotechnology). Sixteen hours posttransfection, the medium was replaced with DMEM with antibiotics, and the cells were subsequently used for luciferase experiments or expression analysis. The effect of YY1 siRNA on APH-1A promoter activity was measured by luciferase reporter assays. APH-1A mRNA and protein levels were analyzed by real-time PCR and western blot, respectively.
Protein levels of APH-1A expression in the siRNA-transfected cells were assessed by Western blot. Briefly, transfected cells were washed three times by PBS and then scraped off in cell lysis buffer. After cells were lysed on ice for 30 min, the cell lysates were cleared by centrifugation at 22 000 g for 30 min at 4 °C. Equal amount of total protein (30 μg) from each sample was loaded onto a 10% Tris-glycine SDS–PAGE gel and then electroblotted onto polyvinylidene difluoride (PVDF) membranes (Millipore Co., Billerica, MA, USA). After blocking with 5% nonfat dry milk, the membranes were incubated with primary antibodies (mouse anti-APH-1A 1:1000; mouse anti-β actin 1:2000; Santa Cruz Biotechnology) and subsequently with horseradish peroxidase (HRP)-conjugated anti-mouse IgG (1:2000; Santa Cruz Biotechnology). Antibody binding was visualized using the enhanced chemiluminescence (ECL) Western blot detection reagent (Amersham Biosciences Ltd, Hong Kong, China), captured on film, and analyzed by Glyko bandscan 4.30 software (Microsoftware Inc., Seattle, WA, USA).
Hardy–Weinberg equilibrium was tested in http://analysis.bio-x.cn/myAnalysis.php. Allele and genotype distributions in patients and controls were assessed by χ2-testing. The strength of association between alleles or genotypes and AD was evaluated with the odds ratio (OR) presented with 95% confidence intervals (CI). The results were corrected for age, gender, and APOEε4 status by using logistic regression. Meta-analysis of the two sample sets was performed using the Review Manager 4.2.10 (The Cochrane Collaboration, Oxford, UK). Heterogeneity was estimated using the Cochrane Q statistic. A fixed-effect model was used for meta-analysis, where the hypothesis of homogeneity was not rejected (P >0.05). Otherwise, a random-effects model was used. Combined OR and 95% CI values were estimated using the Mantel–Haenszel method in the fixed-effect model and the DerSimonian and Laid method in the random-effects model. Group comparisons for the dual-luciferase reporter assay and expression were analyzed using an unpaired two-tailed t test, or a one-way anova when there were more than two groups. All data were analyzed using Statistic Package for the Social Science (spss) version 13.0, and all tests were two-sided. Differences between groups were considered statistically significant when P value was <0.05.
We gratefully acknowledge the technical support and helpful discussions with our colleagues and collaborators. This work was supported by the National Natural Science Key Foundation of China (30830045), National Key Technology R&D Program in the Eleventh Five-year Plan Period (2006BAI02B01), China National Nature Science Foundation (81000472), and Beijing Natural Science Foundation (7102071).