|Home | About | Journals | Submit | Contact Us | Français|
Recent transcriptomics efforts have revealed that numerous protein-coding messenger RNAs have natural antisense transcript partners, most of which seem to be noncoding RNAs. Here we identify a conserved noncoding antisense transcript for β-secretase-1 (BACE1), a crucial enzyme in Alzheimer’s disease pathophysiology. The BACE1-antisense transcript (BACE1-AS) regulates BACE1 mRNA and subsequently BACE1 protein expression in vitro and in vivo. It seems that the argument for concordant regulation can only be made in the experiments with the siRNA against BACE1-AS. This convention has been followed throughout the manuscript. Please check carefully.]. Upon exposure to various cell stressors including amyloid-β 1–42 (Aβ 1–42), expression of BACE1-AS becomes elevated, increasing BACE1 mRNA stability and generating additional Aβ 1–42 through a post-transcriptional feed-forward mechanism. BACE1-AS concentrations were elevated in subjects with Alzheimer’s disease as well as in amyloid precursor protein transgenic mice. These data show that BACE1 mRNA expression is under the control of a regulatory noncoding RNA that may drive Alzheimer’s disease–associated pathophysiology. In summary, we report that a long noncoding RNA is directly implicated in the increased abundance of Aβ 1–42 in Alzheimer’s disease.
Sequential cleavage of amyloid precursor protein (APP) by BACE1, the β-site cleaving enzyme essential for Aβ 1–42 and amyloid-β 1–40 (Aβ 1–40) biosynthesis1, and γ-secretase initiates the ‘amyloid cascade’ that is central to Alzheimer’s disease pathophysiology2,3. Oligomers of Aβ 1–42 produced by BACE1 influence key aspects of Alzheimer’s disease4-9. Studies have revealed elevated brain BACE1 concentrations in subjects with Alzheimer’s disease compared with normal controls10-15. However, controversy exists concerning the extent of BACE1 upregulation and whether this upregulation involves BACE1 mRNA or protein16-18.
Loss of BACE1 results in numerous behavioral and physiological deficits, including memory loss19, emotional deficits20, myelination defects in peripheral nerves21,22 and loss of synaptic plasticity20. Thus, the subtle but crucial boundaries between BACE1 physiology and pathology indicate that BACE1 expression must be tightly regulated, allowing the enzyme to perform its physiological functions while avoiding the serious consequences of over- or underexpression.
Here we report that BACE1-AS, a natural antisense transcript, plays a part in determining BACE1 expression. BACE1-AS rapidly and reversibly upregulates BACE1 levels in response to a variety of stresses, including Aβ 1–42 exposure. Furthermore, we show elevated BACE1-AS in several brain regions of individuals with Alzheimer’s disease. These data suggest that this previously unexamined noncoding RNA has a role in regulating BACE1 and in driving Alzheimer’s disease pathology.
BACE1-AS was originally identified through the FANTOM large-scale transcriptomics efforts as one member of some 1,000 sense-antisense pairs conserved between human and mouse23. BACE1-AS is a conserved ~2-kb RNA transcribed from the positive strand of chromosome 11, on the opposite strand of the BACE1 locus (11q 23.3), including 104 nucleotides of full complementarity to exon 6 of human BACE1 mRNA (Fig. 1a). We performed rapid amplification of cDNA ends (RACE) for directional sequencing of 5′ and 3′ ends and identified two splice variants for human and mouse BACE1-AS that overlap the BACE1 sense transcript (Fig. 1b,c). We found a poly-A tail and cap structure in both human and mouse sequences, indicating that BACE1-AS is a fully processed transcript of RNA polymerase II. However, there is no apparent open reading frame. Sequencing data for human and mouse BACE1-AS are illustrated in Supplementary Data 1 and 2 online.
BACE1 mRNA expression levels were 25–75% greater than BACE1-AS transcript levels across all samples examined from various tissues and cell lines, in contrast to the relative concentrations in Alzheimer’s disease brain (described below). We also observed Bace1 and Bace1-AS transcripts in various regions of the mouse brain. (Supplementary Figs. 1 and 2 online).
In cultures of glia and cortical neurons of human origin, APP mRNA was about twice as abundant in neuronal cultures compared to glial cultures, whereas BACE1 and BACE1-AS transcripts were two to three times more abundant in the glial cells (Fig. 1d). Northern blot analysis with in vitro–transcribed strand-specific RNA probes confirmed that various human tissues express both BACE1 and BACE1-AS (Fig. 1e). This observation suggests that BACE1 and BACE1-AS expression may be regulated concordantly, as was recently shown for other sense-antisense pairs24,25.
We next investigated whether the BACE1-AS transcript regulates expression of BACE1 mRNA. Unexpectedly, transfection of human SH-SY5Y cells with any one of three distinct small interfering RNA (siRNA) sequences targeting nonoverlapping regions of the BACE1-AS transcript resulted in a statistically significant knockdown of not only the targeted BACE1-AS transcript, but also BACE1 mRNA (Fig. 2a). There are broadly two types of regulation between sense and antisense transcripts. In concordant regulation, like in the case of BACE1-AS, the antisense transcripts change the level of the sense RNA, or corresponding protein levels, in a positive way. In contrast, in discordant regulation, the antisense transcripts have negative (opposing) effects on sense transcripts. We also observed this concordant pattern of regulation in human HEK293T and HEK-SW cells, in which siRNA-mediated knockdown of BACE1-AS resulted in a similar reduction in BACE1 mRNA (data not shown). In a control experiment in HEK-SW cells, the level of BACE2 did not change with BACE1-AS siRNA treatment, lending further support to the specificity of the observed BACE1 regulation by BACE1-AS (Supplementary Fig. 3 online). Furthermore, because three distinct siRNA molecules that exclusively target the BACE1-AS transcript resulted in concomitant reduction of sense BACE1 transcript, it is highly unlikely that the siRNAs acted to knock down BACE1 transcript through a nonspecific (or ‘off-target’) mechanism.
To investigate the effects of long-term siRNAs directed against BACE1-AS on BACE1 expression, we generated stable HEK293T cell lines expressing four distinct short hairpin RNAs (shRNAs) to BACE1 or BACE1-AS transcripts. Three of the BACE1 shRNAs and two of the BACE1-AS shRNAs were functional and induced sustained reduction of BACE1 and BACE1-AS, respectively (Fig. 2b). Cells expressing BACE1-AS shRNA showed reduced BACE1 levels, and vice versa (Fig. 2b).
To examine the dose-response relationship of BACE1 mRNA and BACE1-AS siRNA, we measured the reduction in BACE1 mRNA expression across a range of concentrations of BACE1-AS siRNA (100 pM–20 nM) in HEK-SW cells. The resulting data confirmed that siRNA targeting BACE1-AS reduces the expression of BACE1 mRNA in a concentration-dependent manner (Fig. 2c).
Next, we assessed BACE1 protein abundance after transfection of HEK293T cells with BACE1-AS siRNA or shRNA. Western blotting showed that siRNA against BACE1-AS as well as siRNA against BACE1, (but not control siRNAs), reduced the expression of BACE1 protein (Fig. 2d,e). Thus, BACE1-AS seems to control the expression of BACE1 at both the mRNA and the protein levels. Additionally, for accurate quantification of the effects of siRNA treatments on BACE1 protein expression, we established a method for protein quantification by enzyme complementation assay (ECA, Methods and Supplementary Fig. 4 online). We then measured the changes in BACE1 protein concentration after treatment with siRNAs and shRNAs and observed that siRNAs against either BACE1 or BACE1-AS reduce BACE1 protein abundance by 40–60% (Supplementary Fig. 5 online).
Moreover, overexpression of BACE1-AS led to a fourfold increase in BACE1 mRNA (Fig. 2f). When measured by western blotting, the overexpression of BACE1-AS resulted in increased BACE1 protein abundance (Fig. 2g). These observations further confirm the regulation of BACE1 expression by BACE1-AS, not only at the mRNA but also at the protein level.
We measured by ELISA the amount of Aβ 1–40 and Aβ 1–42 after depletion of BACE1-AS in the HEK-SW cell line that contains mutated APP, swAPP751, so-called Swedish mutation26. We found a reduced concentration of both Aβ 1–40 and Aβ 1–42 in cells treated with BACE1-AS siRNA (Fig. 2h). To rule out possible nonspecific effects of siRNA treatment on APP concentration that may account for the observed decreases in Aβ 1–40 and Aβ 1–42, we assayed the amount of soluble APPα (sAPPα, soluble product of α-secretase in the supernatants of the HEK-SW cells) and total APP abundance by ELISA. Neither sAPPα nor total APP abundance was altered upon BACE1 siRNA or BACE1-AS siRNA treatment (Fig. 2h). These results suggest that BACE1-AS siRNA treatment results in reduced BACE1 protein function without affecting APP or α-secretase products.
The above data support a role for BACE1-AS in regulating BACE1 function in vitro in human cells. Next, we assessed whether orthologous Bace1-AS also regulates Bace1 mRNA and protein in vivo in mouse brain. After 14 d of continuous siRNA infusion, Bace1 mRNA levels were reduced across forebrain regions located adjacent to the third ventricle in mice treated with either Bace1 siRNA or Bace1-AS siRNA, compared to levels unaltered by control siRNA (Fig. 3a–d). Bace1 and Bace1-AS transcripts were unaltered in the cerebellum, a structure that is spatially restricted from the third ventricle, of siRNA-treated mice, consistent with previous work that indicates limited penetration of pump-mediated infusion of siRNA into the brain27 (Fig. 3e). We also measured the amount of Bace1 protein in the ventral hippocampus and the cerebellum by western blotting after siRNA treatment. We found that both Bace1 siRNA and Bace1-AS siRNA treatment resulted in reduced Bace1 protein abundance (Fig. 3f). Our in vivo findings agree with the in vitro data described above and indicate that reduced Bace1-AS expression results in reduction of Bace1 mRNA and protein expression in vivo.
Different cell stressors have long been implicated in the pathogenesis of Alzheimer’s disease12,28. We exposed HEK-SW cells to hyperthermia, serum starvation, staurosporine, Aβ 1–42, Aβ 1–40, hydrogen peroxide (H2O2) or high glucose for 12 h. We found that exposure of the cells to high temperature, serum starvation, Aβ 1–42, H2O2 or high glucose resulted in a ~30–130% increase in BACE1-AS levels and a ~20–60% increase in BACE1 mRNA levels (Fig. 4a). Serum starvation generated the strongest response, whereas Aβ 1–40 and staurosporine exposure did not significantly alter BACE1-AS expression levels(Fig. 4a). These results suggest that many, but not all, cell stressors can contribute to the pathogenesis of Alzheimer’s disease by altering BACE1-AS expression and subsequently BACE1 enzyme activity.
Accumulating evidence describes Aβ 1–42 itself as a potent cell stressor10,29-32. To test the hypothesis that Aβ 1–42 increases BACE1 expression by a BACE1-AS dependent mechanism, we exposed SH-SY5Y cells for 2 h to conditioned media from CHO-7PA2 cells, which overexpress APP and generate Aβ 1–42 dimers and oligomers33. Exposure of the SH-SY5Y cells to conditioned media from CHO-7PA2 cells, but not conditioned media from control parental CHO cells resulted in an increase in cytoplasmic concentrations of BACE1-AS transcript (Fig. 4b). We obtained similar results when incubating SH-SY5Y cells with synthetic Aβ 1–42 (1 μM for 2 h; Fig. 4c). Removal of the cell stressors normalized BACE1-AS expression patterns. Using an ECA, we found that synthetic Aβ 1–42 (1 μM for 12 h) elicited an increase in BACE1 protein abundance as well (Fig. 4d). Taken together, the above data indicate that cell stress increases BACE1-AS levels, which in turn increases BACE1 levels; this may result in an increase in APP processing and Aβ 1–42 production. Subsequently, increased Aβ 1–42 levels can further increase BACE1-AS expression, driving the APP processing cascade in a feed-forward manner.
We used an RNase protection assay (RPA) on RNA from SH-SY5Y cells to test the possibility of RNA duplex formation (Supplementary Methods). RT-PCR data showed that the overlapping part of both transcripts was protected from degradation, indicating that BACE1 and BACE1-AS indeed form a RNA duplex (Fig. 5a). We also validated the RPA data on a 10% Tris-borate-EDTA–urea gel using radiolabeled BACE1-AS probes (data not shown).
RNA duplex formation may act to alter the secondary or tertiary structure of BACE1 and thereby increase its stability. We assessed the stability of BACE1 and BACE1-AS transcripts by blocking new RNA synthesis with α-amanitin and measuring the loss of BACE1, BACE1-AS, β-actin (ACTB) and 18s RNA over a 24-h period. We found that BACE1-AS had a shorter half-life than BACE1 mRNA (8.5 h versus 17.5 h Fig. 5b). 18s ribosomal RNA, which is a product of RNA polymerase I, was not affected by α-amanitin treatment (Fig. 5b). In a cell line that constitutively expresses BACE1-AS shRNA and thereby has depleted BACE1-AS levels, we found decreased stability of BACE1 mRNA compared with cells transfected with a control shRNA (Fig. 5c). Conversely, cells that overexpress BACE1-AS showed increased stability of BACE1 (Fig. 5d). Collectively, our data demonstrate that BACE1-AS increases the stability of BACE1 mRNA.
Elevated BACE1-AS concentrations may facilitate increased BACE1 activity and disease progression in the brains of human subjects with Alzheimer’s disease. To examine this question, we assessed BACE1-AS and BACE1 mRNA abundance in RNA samples prepared from parietal lobes and cerebellum from five postmortem brains of human subjects with Alzheimer’s disease and from five age- and sex-matched control brains. In the Alzheimer’s disease samples, the relative quantity of BACE1-AS transcript was increased by two to three times, along with a smaller increase in BACE1 transcript (Fig. 6a and Supplementary Fig. 6a online). In a separate group of 35 subjects with Alzheimer’s disease and 35 age- and sex-matched controls34, we examined RNA samples derived from cerebellum (25 Alzheimer’s disease samples and 21 control samples), hippocampus (13 Alzheimer’s disease samples and 11 control samples), entorhinal cortex (13 Alzheimer’s disease samples and 11 control samples) and superior frontal gyrus (16 Alzheimer’s disease samples and 17 control samples). The BACE1-AS transcript concentrations were elevated in subjects with Alzheimer’s disease by up to sixfold, with an average elevation of about twofold across all brain regions (Fig. 6b–d and Supplementary Fig. 6b,c). We detected a smaller (~30%) increase in BACE1 mRNA concentrations in these subjects compared to their matched controls (Fig. 6a,b) [AU: Figure callout or ‘data not shown’?]. Taken together, these results support our hypothesis that increases in BACE1-AS expression, probably related to cell stressors, drives upregulation of BACE1 mRNA and protein level, thereby facilitating Aβ 1–42 biosynthesis in human Alzheimer’s disease brain.
BACE1-AS may have utility as a new biomarker of Alzheimer’s disease35. To this end, we calculated the ratio of BACE1-AS relative to BACE1 and ACTB mRNA in different brain regions of control subjects and subjects with Alzheimer’s disease. We found that the BACE1-AS to ACTB ratio was increased in various brain regions in subjects with Alzheimer’s disease as compared to control individuals (Fig. 6e). A smaller increase in the BACE1-AS to BACE1 ratio was also observed in the brains of individuals with Alzheimer’s disease (Fig. 6f). These data demonstrate that the ratio between BACE1-AS and other RNA transcripts, including BACE1, could potentially be used as a biomarker of Alzheimer’s disease.
Tg19959 mice, considered a mouse model of Alzheimer’s disease, overexpress a doubly mutated human APP (APP-tg19959)36 and consequently have increased levels of Aβ 1–42 (ref. 37). Samples from whole brains excised from four six-week-old male mice had increased (~300-fold) levels of Aβ 1–42 compared with samples from matched wild-type controls, as measured by homogeneous time resolved fluorescence (HTRF) assay (Fig. 6g and Supplementary Methods). Expression of the Bace1-AS transcript was increased by about 45%, and Bace1 mRNA expression was increased by about 25% in the brains of the APP-tg19959 mice compared with wild-type control mice (Fig. 6h), similar to the measurements in the human samples.
The contrast between BACE1’s essential role in cognitive, emotional and synaptic functions19,20 and its pathophysiological dysregulation in Alzheimer’s disease38,39 highlights the regulatory complexity of this protein. Owing to the consequences of its dysregulation, BACE1 gene expression must normally maintain tight robust regulatory control.
In this study, we have characterized a conserved noncoding antisense transcript for BACE1, called BACE1-AS, which functions as a regulator of BACE1 gene expression. We present data showing that BACE1-AS is widely co-expressed with BACE1 in cell lines, tissues and Alzheimer’s disease–sensitive brain regions and that it regulates BACE1 expression in vitro and in vivo. We found that selective siRNA targeting of the nonoverlapping regions of the BACE1-AS resulted in reduction of BACE1 mRNA and protein abundance in vitro. Administration of siRNA that selectively targeted either Bace1 or Bace1-AS into mouse brains reduced the levels of both transcripts, indicating that this concordant regulation also occurs in vivo.
In addition, we have shown that alterations in BACE1-AS RNA concentrations can alter Aβ 1–40 and Aβ 1–42 production. Considering the narrow window between essential levels and excessive levels (as in Alzheimer’s disease) of BACE1 protein, we believe that neuronal cells must maintain precise physiological regulation of BACE1 expression by using both pre- and post-transcriptional regulatory mechanisms. The RNA transcript, BACE1-AS, seems to function as a regulatory component of this machinery.
Because BACE1-AS regulates BACE1 expression in vivo, we propose that the elevation of BACE1-AS, resulting from the actions of Alzheimer’s disease–related cell stressors, forms a basis for a deleterious feed-forward cycle of Alzheimer’s disease progression. Even small changes in BACE1 activity may lead to a long-lasting and chronic process of Aβ 1–42 accumulation in the Alzheimer’s disease brain38,40. Our current findings provide further evidence for a feed-forward mechanism of stress-dependent and activity-dependent41 Aβ 1–42 production. Recent studies have shown that amyloid plaques induce elevation of BACE1 protein expression in adjacent neurons by a post-transcriptional mechanism10. This finding is consistent with the present data in which Aβ 1–42 was shown to induce increased levels of BACE1-AS, thereby driving BACE1-mediated APP processing and further accumulation of Aβ 1–42. In support of the above interpretation, we found that two independent sets of human Alzheimer’s disease brain samples as well as an animal model of Alzheimer’s disease express elevated levels of both BACE1-AS transcript and, to a lesser degree, BACE1 sense transcript. In contrast to the downregulation of most transcripts reported to date in Alzheimer’s disease brains42,43, BACE1-AS upregulation may be the driving force behind Alzheimer’s disease–related BACE1 dysregulation15,44,45. Thus, our results implicate a noncoding RNA in the control of gene expression central to the delicate balance between healthy stress response and the pathophysiological β-amyloid cascade.
Aβ 1–42 induces synaptic depression by triggering endocytosis of glutamatergic N-methyl d-aspartate receptors from the post-synaptic membrane6. Synaptic activity–dependent production of Aβ 1–42 achieves this depressive effect of N-methyl d-aspartate receptor endocytosis by a series of common mechanisms that implicate Aβ 1–42 in the establishment of some forms of long-term depression46. Although they contribute to the depth and richness of mammalian memory, if not precisely controlled, these mechanisms may lead to chronic neuronal stress and the onset of Alzheimer’s disease. We have previously speculated that noncoding RNAs may be required for some forms of long-term depression47, and BACE1-AS could well be involved in such a function.
Treatment with BACE1-AS siRNA may achieve a preferential reduction of stress-induced increases in BACE1 expression without disturbing physiologically essential basal expression levels. Thus, we propose that BACE1-AS could potentially constitute a drug target candidate well suited to mediate the transition between the essential physiological functions of BACE1 and its pathological dysregulation in the chronically stressed setting of early Alzheimer’s disease48. Our in vivo experiments using infusion of unmodified synthetic siRNA over an extended period of time in experimental mice support the validity of an siRNA approach to decrease BACE1 expression, perhaps in humans as well. A recent technological breakthrough suggests that systemic administration of modified siRNA may cross the blood-brain barrier and thereby target RNA transcripts in the brain49. Alternatively, proteins involved in BACE1-AS localization or turnover could serve as potential targets for therapeutic interventions.
ECA is a technology developed by DiscoveRx that allows for the measurement of changes in protein abundance (Supplementary Fig. 2). We cloned the cDNA of BACE1 into a pCMV-ProLabel vector upstream of the ProLabel. We transfected vector into HEK293T cells to produce a fusion protein (BACE1 and the enzyme donor fragment of β-galactosidase) and made a stable cell line, which we called C3. In our experiments, we treated HEK293T cells with Aβ 1–42 peptides for 12 h and then added the lysis buffer (DiscoveRx), which includes the enzyme acceptor fragment of β-galactosidase. When the two fragments of the β-galactosidase combine in solution, the enzyme becomes active and hydrolyzes a substrate that produces a chemiluminescent signal. The strength of this signal is proportional to the protein being produced (in this case, BACE1). In a separate experiment, we transfected the stable cell line C3 overexpressing BACE1 with siRNA or shRNA against BACE1, BACE1-AS or control siRNA and measured protein expression 72 h later with this methodology. We plotted data as a percentage of control siRNA.
The first set of human brain samples was prepared at the USC Alzheimer’s Disease Research Center. The USC Alzheimer’s Disease Research Center obtained informed consent from all subjects and the USC Institutional Review Board approved the use of the human tissue. RNA was extracted from parietal lobes and cerebellum of postmortem brains of five subjects with Alzheimer’s disease and five matched controls. The average age of subjects with Alzheimer’s disease was 85 years (range 75–92 years) and 91.8 years (90–95 years) for controls. The postmortem interval ranged from 3.75–10.1 h with a mean of 5.87 h. We treated RNA samples with DNase and purified them with RNeasy mini columns (QIAGEN). We prepared cDNA from 400 ng of RNA samples and used RT-PCR for relative quantification of different transcripts. The second set of human brain samples was prepared from rapid autopsy brain tissue that had been obtained from J. Rogers (Sun Health Research Institute); all enrolled subjects or legal representatives had signed a Sun Health Research Institutional Review Board–approved informed consent form allowing both clinical assessments during life and several options for brain and bodily organ donation after death. These cases included 35 autopsy-confirmed cases of Alzheimer’s disease with an average age of 81.8 years (range 64–92 years) and 35 controls with an average age of 72.3 years (range 53–91 years). The postmortem interval ranged from 1.25–5 h with a mean of 2.5 h. The average duration of disease in the subjects with Alzheimer’s disease was 9.2 years. Total RNA was isolated via CsCl purification from tissue dissected from specific regions of brain. Although not all regions were available from all cases, we examined a total of 128 RNA samples from superior frontal gyrus, entorhinal cortex, hippocampus and cerebellum for BACE1 and BACE1-AS expression by RT- PCR.
We obtained approval for mouse studies from the Institutional Animal Care and Use Committee at The Scripps Research Institute.
We used 18 six-month-old male mice for in vivo siRNA infusion experiments. We prepared mice with chronic indwelling cannulae in the dorsal third ventricle implanted subcutaneously with osmotic minipumps that delivered continuous infusions (0.25 μl/h) of synthetic unmodified siRNA directed against Bace1, Bace1-AS or control siRNA (previously shown to have no effects on the expression of human and mouse genes) at a dose of 0.4 mg/d for 2 weeks. We connected tubing to the exit port of the osmotic minipump and tunneled it subcutaneously to the indwelling cannula, such that siRNAs were delivered directly into the brain.
Pump-mediated infusion of siRNA was previously shown to significantly and specifically knock down expression of targeted mRNAs in the brain, but with a limited tissue penetration27. Indeed, RNA knockdown upon ventricular infusion of siRNAs for 14 consecutive d was usually obtained in brain regions immediately adjacent to the ventricle, with diminishing effects of the siRNA as the distance from the ventricle increased.
We excised five tissues from each mouse for RNA quantitative measurement—the dorsal hippocampus, ventral hippocampus, cortex, dorsal striatum and cerebellum. RNA extraction is described in Supplementary Methods.
Tg19959 mice were produced by pronuclear microinjection of (FVB × 129S6F1) embryos with a cosmid insert containing human APP with two familial Alzheimer’s disease mutations (KM670/671NL and V717F) under the control of the hamster PrP promoter. We euthanized four Tg19959 mice and four control male littermates at 6 weeks old. We used brain tissues for RNA measurements and Aβ 1–42 detection by HTRF. In a separate experiment, we euthanized three wild-type male mice and excised their tissues for expression profiling of Bace1 and Bace1-AS by RT-PCR.
We thank J. Rogers (Sun Health Research Institute) for autopsy brain tissue. We are grateful to M. Leissring (Mayo Clinic, Jacksonville) for helpful discussions and for kindly providing cell lines and APP-tg19959 mouse materials. We also thank D. Willoughby for his help in RNA purification from human Alzheimer’s disease samples. S. Brothers provided valuable help in manuscript preparation. M.A.Faghihi. is partly supported by a scholarship from the Ahwaz University of Medical Sciences, Ministry of Health I.R. Iran. This study has been supported in part by the US National Institutes of Health (AG 029290).