|Home | About | Journals | Submit | Contact Us | Français|
The clinical hallmark of Alzheimer's disease (AD) is impairment of cognition associated with loss of synapses, accumulation of amyloid β (Aβ) both within neurons and as extracellular deposits, and neurofibrillary degeneration composed of phospho-tau. Neurons in the hippocampus are among those that are most vulnerable. The purpose of this study was to investigate the expression of genes associated with cognition, synapse, and mitochondrial function in hippocampal neurons of AD compared to normal individuals. Neurons from the hippocampus with intraneuronal Aβ immunoreactivity were captured with laser microdissection; RNA was extracted; and levels of brain-derived neurotrophic factor (BDNF), TrkB (BDNF receptor), dynamin-1 (DYN), and cytochrome C oxidase subunit II (COX2) were assessed with quantitative real time-polymerase chain reaction. We found no significant differences in the expression of these genes in AD between neurons associated with Aβ compared to those lacking Aβ immunoreactivity. Double immunofluorescence microscopy showed the number of hippocampal neurons coexpressing Aβ or phospho-tau and either BDNF, TrkB, or DYN was similar in AD and controls. Our results suggest that intraneuronal Aβ or phospho-tau do not have obligatory effects on reducing the expression of genes important for memory and cognition in hippocampus of AD.
The clinical hallmark of Alzheimer's disease (AD) is impairment of memory and cognition that correlates with accumulation of Aβ within neurons as well as in extracellular deposits referred to as senile plaques [1–3]. Aβ is derived from amyloid precursor protein (APP) by regulated intramembranous proteolysis and eventually is broken down or eliminated . In AD, these processes are disrupted. Accumulation of soluble forms of Aβ in the hippocampus and amygdala appears to be one of the earliest events that lead to cognitive impairment [5, 6]. More recently, injection of a natural soluble form of Aβ into rat brain rapidly and potently impaired cognitive function without inducing permanent neurological deficits . These results led to the hypothesis that low amounts of diffusible forms of Aβ can directly reduce expression of neuronal genes involved in memory and cognition.
Neuronal synaptic dysfunction is a significant factor contributing to memory loss in AD [8–12]. Evidence suggests that Aβ may affect hippocampal synaptic efficacy before pathological structural degeneration [13, 14]. For example, treatment of hippocampal neurons in vitro with Aβ results in decreases in expression of dynamin-1 (DYN), a protein important in synaptic function . Similarly, mitochondrial genes are up-regulated in APP transgenic mice because of oxidative damage at the earliest stage, indicating impairment of mitochondrial energy metabolism by Aβ . Aβ peptides also cause leaks in the mitochondrial membrane, resulting in the release of cytochrome c, activation of caspase-3, and apoptotic death of neurons [17, 18]. It is possible that intraneuronal Aβ has direct effects on expression of synaptic or mitochondrial genes as an early event in the molecular pathogenesis of AD [3, 13, 16, 19, 20].
We hypothesize that if intraneuronal Aβ directly reduces the expression of neuronal genes associated with neuronal growth and survival, as well as synaptic and mitochondrial functions, hippocampal neurons containing Aβ immunoreactivity will have lower expression of these genes compared to neurons not associated with intraneuronal Aβ. Brain-derived neurotrophic factor (BDNF) and its receptor TrkB are representative genes important for memory and cognition that were analyzed in this study [21–23]. We also analyzed DYN, important for synaptic vesicle recycling and acquisition of memory, and cytochrome C oxidase subunit II (COX2), important for mitochondria function [24, 25]. It is clear from previous studies that the expression of BDNF, TrkB, and DYN is lower in brains of AD compared to controls [26–31]. Cytochrome oxidase activity is decreased, but the expression of mitochondrial gene, such as COX2, is increased, possibly due to compensatory effects on surviving neurons [19, 25, 32, 33]. At present, it is not clear to what extent the changes in the expression of these genes are due to direct effects of intraneuronal Aβ.
In the present study, we used laser capture microdissection to isolate hippocampal neurons associated with intraneuronal Aβ immunoreactivity, as well as neurons without Aβ, in brains of AD and normal controls and determined the levels of BDNF, DYN, and COX2 by qRT-PCR. We found no significant differences in the expression of these genes in neurons with or without Aβ immunoreactivity in AD. Using double immunofluorescence microscopy (DIF), we confirmed there were hippocampal neurons with Aβ immunoreactivity that express BDNF, TrkB, or DYN in both AD and controls. Similar to previous reports, we found lower levels of BDNF, TrkB, and DYN in AD that were correlated with an overall increase in intraneuronal Aβ.
We then investigated whether hippocampal neurons with phosphorylated tau (P-tau) immunoreactivity express lower levels of BDNF, TrkB, and DYN than neurons without P-tau immunoreactivity. Tau is a microtubule-binding protein that becomes abnormally phosphorylated in AD and accumulates to form neurofibrillary tangles [34, 35]. Our results using DIF were similar to Aβ, with no significant difference in gene expression for BDNF or DYN in neurons with and without P-tau immunoreactivity in AD and controls. We conclude that neither Aβ nor P-tau has obligatory intracellular effects on reducing the expression of genes important for memory and cognition in hippocampus of AD.
Both cryosections and paraffin-embedded hippocampal tissue sections (10 μm thick) were obtained from AD and age-matched controls. At the time of autopsy 0.5-cm thick pieces of hippocampus were placed in OCT medium and flash frozen in isopentane chilled with liquid nitrogen. Sections were cut at 10 μm thickness and stored at −80°C before analysis. Paraffin sections were obtained from additional AD cases and controls that had been fixed for 10–14 days in neutral buffered formaldehyde before embedding. All cases were confirmed by standardized clinical and neuropathologic methods that included quantitative assessment of senile plaques and neurofibrillary tangles with thioflavin-S fluorescent microscopy in multiple cortical sections and the hippocampus. Tissue was obtained from Mayo Clinic, Jacksonville, FL. Pathological information on the cases is summarized in Table 1. The time interval from death to freezing or fixation of tissue was between 2 and 11 hours.
All procedures were carried out in an RNase-free environment, and all solutions contained placental RNase inhibitor (40 U/100μl) from Invitrogen (Carlsbad, CA). Slides were processed singly with rapid staining and washed briefly with phosphate-buffered saline (PBS) after each step. For immunostaining prior to laser capture, we followed the procedures of Fend F, et al. and D'Andrea, et al [36, 37]. The procedure is as follows: cryosections of AD or controls were thawed, fixed with 75% ethanol for 30 sec, and then immunostained for 5 min at room temperature for Aβ using a rabbit polyclonal antibody (0.83 μg/μl) from Invitrogen diluted to 1:625 with 1.5% goat serum in PBS. Labeled cells were identified by incubating for 2 min at 37°C with fluorescein isothiocyanate (FITC)-labeled swine anti-rabbit secondary antibody (DAKO, Carpinteria, CA) diluted to 1:10 with 1.5% goat serum in PBS. Slides were pretreated with 0.05% pontamine and 0.1% Triton X-100 to minimize background. Subsequently, neurons were morphologically identified by staining with 1% Nissl (Sigma, St. Louis, MO; ref. 38) for 30 sec, dehydrated with xylene, and stored in a vacuum desiccator until microdissection. Sections were microdissected using a Leica laser capture microdissection (LCM) unit at the Imaging Laboratory, University of Miami Miller School of Medicine (Miami, FL). From an AD case, Aβ positive and negative Nissl stained neurons were collected into separate tubes (Fig. 1). Neurons from normal controls often displayed only Nissl staining. On average, ~500 neurons of each category were collected per slide for AD and ~1000 neurons per slide for the controls. Neurons from six slides for each case were pooled. Three AD cases (AD1, AD2, AD3) and four controls (C1, C2, C3, C4) were processed (Table 1).
RNA from pooled neurons for each case was extracted using an Absolutely RNA Nano kit from Stratagene (La Jolla, CA) and treated with RNAse-free DNase I (Invitrogen) to remove DNA. The concentration for each RNA preparation was determined using the Ribogreen kit (Invitrogen). RNA was asymmetrically amplified through two cycles using a Message Amp aRNA kit from Ambion (Austin, TX). In addition, total RNA from an age-matched normal case from the hippocampus tissue was obtained from Ambion.
Four µg or less of RNA isolated from LCM neurons or hippocampus tissue was reverse transcribed to cDNA with random primers using the First Strand cDNA Synthesis kit for RT-PCR from Roche Applied Sciences (Indianapolis, IN) in a thermo-cycler GeneAmp 9600 PCR from Perkin-Elmer (Waltham, MA). Primer-pair sequences specific for BDNF (Qiagen, Chatsworth, CA), DYN (Operon Huntsville, AL), and COX2 (TIB Mol Biol, Adelphia, NJ) were designed using published literature or the Primer 3 program (http://frodo.wi.mit.edu) (Table 2). Ribosome elongation factor 1α (EF1α) was used as the reference gene for the calculation of Ro  (Table 2). Quantitative RT-PCR was performed using the Light Cycler DNA Master SYBR Green I kit with varying amounts of cDNA (1 to 200 ng) and a Light Cycler system from Roche Applied Sciences (Indianapolis, IN). The qRT-PCR and melting point (Mp) profiles were obtained. Each product was confirmed for the expected size by agarose gel electrophoresis. For EF1α and COX2, PCR was started with denaturation at 95°C, 30 sec, followed by amplification for 45 cycles at 95°C (0 sec, 20°C/sec), annealing at 50°C (10 sec, 5°C/sec), and elongation at 72°C (10 sec, 2°C/sec),and linked to detection at 80°C (0 sec, 20°C/sec). For BDNF and DYN, target genes with small copy number, PCR was started with a two-step denaturation at 50°C for 2 min, 95°C for 2 min, followed by amplification for 45 cycles at 95°C 15 sec, 58°C 30 sec, 72°C 45sec, and detection at 80°C . The qRT-PCR profile was obtained by the instrument software using the fluorescent intensity at the detection temperature for each cycle. Negative controls were without reverse transcriptase (RT), resulting in no PCR product. The EF1α reference gene was always analyzed using the same conditions as the target gene for calculation of Ro. Age-matched hippocampus RNA was also included in each run and used as a reference (Ro=1). Each sample was analyzed at least three times from at least three independent experiments.
Ro, the initial ratio of target RNA of a specimen to E1Fα RNA, was obtained by the equation derived by Pfaffl . Ro for each specimen was averaged from at least three independent experiments. The amplification efficiency (E), whose log is the slope for the logarithmic phase of each amplification curve, was obtained by the PSI-plot software from Poly Software International (Salt Lake City, UT) or by the Microsoft Excel program . Crossing point (Cp) or the threshold of logarithmic amplification was obtained as the second derivative maximum using the instrument software. Statistical analyses as described by Fujimura, et al.  showed that the ratios of target genes to the reference gene EF1α (Ro) for each set of specimens were independent of cDNA concentrations. Therefore, the significant differences observed in Ro between specimens were due to differences in amounts of target gene RNA. For each target gene, the analysis of variance was performed to compare Ro values between AD (n=3) and controls (n=4). Significance level was set at 0.05 for all tests. A post-hoc power analysis showed that with a two-sided 0.05 significance level and assuming no interclass correlation, there is an 80 percent and 88 percent power to detect an effect size of 1.75 and 2.00, respectively.
For DIF, additional paraffin-embedded hippocampal sections from AD and controls were used (Table 1). Endogenous peroxidase activity was blocked using 3% H2O2 in methanol for 5 min. For antigen retrieval, sections were incubated in hot 10 mM citrate buffer (pH 6.0) for 20 min. Immunofluorescence for Aβ (Abcam, Cambridge, MA; ab10148), P-tau (clone AT8; Pierce Biotechnology, Rockford, Il), BDNF (Santa Cruz Biotechnology, Santa Cruz, CA; H-117), TrkB (Cell Signaling, Beverly, MA; clone 80G2), and DYN (BD Biosciences, San Jose, CA; clone 41) was done using a 1:50 dilution of rabbit polyclonal or mouse monoclonal antibodies for 30 min at room temperature. After incubation with antibodies to either Aβ or P-tau, antibody binding was detected with biotinylated anti-rabbit or mouse IgG (1/200 dilution; 20 min at room temperature; Vector Laboratories, Burlingame, CA), a 1:300 dilution of FITC-avidin DCS (Vector Laboratories) for 5 min, followed by avidin/biotin blocking for 15 min, and then immunostaining with BDNF, TrkB, or DYN and detection with biotinylated anti-rabbit or mouse IgG and Texas Red DCS for 5 min. Psuedocolor images were captured using a Nikon fluorescence microscope with FITC and Texas Red filters using IP Lab software (Fairfax, VA) and merged using Adobe Photoshop (San Jose, CA). The number of Aβ, P-tau, BDNF, TrkB, and DYN positive cells from 3–16 high-power fields (x200) were chosen from the pyramidal layer in areas with the highest density of immunostaining for three AD (AD5, AD6, AD7) and three controls (C5, C6, C7). The number of cells coexpressing Aβ/TrkB, Aβ/DYN, P-tau/BDNF, and P-tau/DYN was determined by counting yellow cells from the merged images. For negative controls, we used the same concentration of mouse or rabbit IgG (Santa Cruz Biotechnology) instead of specific primary antibodies, resulting in lack of immunostaining.
Total human hippocampus tissue lysates from normal and AD and from normal prostate was purchased from ProSci, Inc. (Poway, CA). After separation of 5!10 μ-g protein by SDS-PAGE, proteins were transferred by electrophoresis to Immobilon-P membrane and incubated in 5% nonfat dry milk, TBS, and 0.1% Tween-20 for 1 hour. The same antibodies used for DIF were diluted 1/1,000 in 5% nonfat dry milk, TBS, and 0.1% Tween-20 and incubated overnight at 4EC. Membranes were washed in TBS and 0.1% Tween-20 and incubated with the appropriate horseradish peroxidase-conjugated secondary antibody (1/2,000 dilution; Santa Cruz) for 1 hour, washed in TBS and 0.1% Tween-20, and analyzed by exposure to X-ray film using enhanced chemiluminescence plus (ECL plus, GE Healthcare Bio-Sciences Corp., Piscataway, NJ). Staining of total protein with Coomassie blue was used as protein loading controls. X-ray films were scanned using an Epson Perfection 2450 Photo scanner.
We hypothesized that Aβ-associated neurons would have lower expression levels of genes essential for memory, as well as synaptic and mitochondrial functions, compared to Aβ-free neurons. To test this hypothesis, we obtained frozen sections of hippocampal tissue from AD and controls (Table 1). Aβ-positive and Aβ-negative Nissl-positive neurons were obtained by LCM from three AD and four control cases and analyzed by qRT-PCR for the expression of BDNF, DYN, and COX2 (Figs. 1, ,2).2). The results showed that for each AD case, there were no significant differences in the levels of BDNF, DYN, or COX2 between Aβ-positive and Aβ-negative neurons (Fig. 2). In addition, the changes in expression of COX2 and DYN were not necessarily linked to changes in expression of BDNF. For BDNF, a gene essential for memory and cognition, the expression in Nissl-positive neurons of the controls was much higher (Ro = 10) compared to total hippocampal RNA, while in AD the expression in Nissl-positive neurons was much lower (Ro = 0.5) (Fig. 2A). For TrkB, a receptor for BDNF, the expression in AD cases was also very low (Ro ≤ 0.01) (data not shown). For DYN, a gene essential for synapses, the expression was similar between AD and controls, but much lower than in total hippocampal RNA (Ro ≤ 0.1) (Fig. 2B). For mitochondrial COX2, the expression was similar in controls compared to total hippocampal RNA (Ro = 1), with substantial variation in AD (Fig. 2C). Thus the changes in the expression of these three target genes did not appear to be linked. We conclude that there may not always be a direct relationship between memory and cognitive impairment and mitochondrial or synaptic dysfunction in AD, contrary to the expectation from the literature [3, 6, 8–13, 18–20].
We used DIF to determine if hippocampal neurons with Aβ immunoreactivity that expressed BDNF, TrkB, or DYN were present in both AD and controls. As expected, we found more neurons with intracellular Aβ immunoreactivity in AD compared to controls and more neurons immunoreactive for BDNF, TrkB, and DYN in controls compared to AD (Figs. 3, ,4,4, ,7).7). The number of hippocampal neurons double immunostained for Aβ and TrkB or Aβ and DYN (yellow cells in the merged images) was similar in AD and controls (Fig. 7B). These results support the qRT-PCR analysis, in that expression of BDNF and DYN is similarly detected in hippocampal neurons of AD cases that are both Aβ-positive and Aβ-negative.
In addition to Aβ, P-tau, the major structural component of neurofibrillary tangles, has been hypothesized to have an important role in neuronal dysfunction in AD [31, 34, 35]. We used DIF to determine if hippocampal neurons that were immunoreactive for P-tau also expressed BDNF or DYN. Our results showed, as predicted, more neurons with P-tau immunoreactivity in AD compared to normal controls (Figs. 5, ,7A).7A). Similar to the results with Aβ immunoreactivity, the number of hippocampal neurons double-immunolabeled for P-tau and BDNF or for P-tau and DYN (yellow cells in the merged images) was similar in AD and controls (Figs. 5, ,7B).7B). There were fewer double-labeled neurons for P-tau and DYN (0.9–1.3 cells/hpf) compared to Aβ and DYN (2.3–2.5 cells/hpf) in both AD and controls (Fig. 7B); however, the differences were not statistically significant (P = 0.1). The specificity of the antibodies used for DIF were confirmed by Western blot analysis of human hippocampus protein lysates from normal and AD compared to proteins from human prostate (Fig. 6). Overall, these results suggest that intraneuronal accumulation of Aβ and P-tau protein in hippocampal neurons does not have a strong negative effect on expression (mRNA and protein) levels of BDNF, TrkB, or DYN in either AD or controls.
The underlying molecular mechanisms of AD remain to be elucidated. Two prevailing hypotheses for the progressive neurologic deterioration in AD are increases in Aβ in senile plaques and P-tau in neurofibrillary tangles [2, 3, 35, 44]. More recent evidence in transgenic mouse and rat models suggest that the soluble oligomers of Aβ are the toxic species in AD [5–7]. It is not clear in human cases of AD if the increases in Aβ or P-tau in hippocampal neurons directly decreases the expression of genes important for memory and cognition, such as BDNF and TrkB, or genes important in synaptic and mitochondrial function, such as DYN and COX2, respectively. In the present study, our results using LCM and qRT-PCR analysis of RNA showed no differences in the expression of these genes (i.e., BDNF, DYN, and COX2) in neurons of the hippocampus of AD and controls that displayed intraneuronal Aβ immunoreactivity compared to neurons without Aβ immunoreactivity (Fig. 2). Utilizing DIF in AD and controls, we also found a similar number of hippocampal neurons coexpressing Aβ/TrkB, Aβ/DYN, P-tau/BDNF, and P-tau/DYN (Figs. 3–5, ,7B).7B). Therefore, our data suggest that the accumulation of intracellular Aβ and P-tau in hippocampal neurons does not always play a major role in directly reducing expression of BDNF, TrkB, DYN, or COX2.
Data suggesting that Aβ can directly reduce the expression of DYN come from experimental treatment of fetal rat hippocampal neurons grown in culture . A small but significant decrease in DYN protein in hippocampal extracts isolated from the Tg2576 transgenic mouse model of AD has also been demonstrated . These mice contain a double mutation in APP that increases production of Aβ and produces some of the characteristics of AD pathology, in particular, senile plaque formation . The reduction of DYN protein in cultured rat hippocampal neurons and possibly in the Tg2576 transgenic mice may be due to Aβ-mediated increases in calpain degradation of DYN . This potential mechanism of Aβ-mediated increase of calpain degradation of DYN protein may explain why our data in human hippocampal tissue sections from AD show no difference in DYN mRNA levels compared to control cases, whereas our results utilizing DIF show reduced levels of DYN protein in AD compared to controls (Figs. 2, ,3B,3B, ,7A).7A). However, if the Aβ-mediated increase in calpain degradation of DYN protein is a major mechanism, it is difficult to explain why our results show similar numbers of dual-labeled Aβ/DYN hippocampal neurons in AD and controls (Figs. 3B, ,7B7B).
Additional evidence in fetal rat cultured cortical neurons suggests that Aβ can interfere with the ERK and AKT signaling pathways that activate BDNF transcription factors . Another study found that treatment of a human neuroblastoma cell line with soluble Aβ peptides decreases phosphorylated cAMP response element-binding protein (CREB), a regulator of BDNF transcription, and BDNF total mRNA . In the present report, our data are similar to those of other previously published reports showing that BDNF mRNA and protein levels are reduced in AD compared to controls [21–23]. Because our data showed that the expression level of BDNF mRNA was similar in hippocampal neurons with intraneuronal Aβ compared to neurons free of Aβ immunoreactivity (Fig. 2A), it is likely that intraneuronal Aβ-mediated reduction of transcription factors regulating BDNF is not a major feature of AD.
Our results in hippocampal tissue sections are contrary to the hypothesis that intraneuronal Aβ directly reduces the expression of genes important in memory and cognition. However, it is possible that the hippocampus in AD is permeated with soluble oligomeric Aβ peptides that are not visible with immunostaining. We also investigated by DIF whether P-tau was associated with reduced levels of BDNF, TrkB, or DYN in hippocampal neurons of AD and controls. The abnormal accumulation of P-tau is hypothesized to disrupt transport of proteins and growth factors by interfering with microtubule-dependent axoplasmic transport [35, 44]. The results with DIF, using the AT8 monoclonal antibody that specifically recognizes P-tau but not nonphosphorylated tau protein, were similar to the results with DIF for Aβ (Figs. 5, ,7).7). However, in view of our negative findings, future studies with greater sample sizes are warranted.
There are no reports that the overexpression of P-tau has a direct effect in reducing the levels of DYN or BDNF, as has been reported for Aβ in cultured neuronal cells [15, 40, 46]. In contrast, one report shows that BDNF stimulation of differentiated P19 mouse embryonic carcinoma cells results in a rapid decrease in tau phosphorylation, suggesting that BDNF activation of its receptor, TrkB, has a role in lowering P-tau protein . Therefore, one possibility is that reduction of BDNF protein increases P-tau, while P-tau does not directly reduce BDNF. However, if the reduction of BDNF is correlated with increased P-tau, it is difficult to explain why our results show similar numbers of hippocampal neurons coexpressing P-tau and BDNF in both AD and controls (Figs. 5A, ,7B).7B). There were fewer P-tau labeled neurons coexpressing BDNF and these cells were present in the same proportion in the AD and control cases. It would be of interest to determine if these cells are the same cells that coexpress Aβ and BDNF.
Perhaps a better way to determine whether Aβ and P-tau protein are maintained at harmless levels in healthy controls is to study animal models. Injection of a natural soluble form of Aβ peptide into rat brain rapidly and potently impairs cognitive function without inducing permanent neurological deficits, although the effect on genes important in cognitive function was not reported . Numerous transgenic mouse models are available that overexpress mutated APP or mutated tau that develops amyloid plaques and neurofibrillary tangles, respectively [48, 49]. The advent of inducible transgenic mouse models that target neuronal cells with APP or tau in a controlled and timed manner may be required to determine whether genes important for memory and cognition, like BDNF, TrkB, and DYN, are reduced [48–51]. Although the neuropathology of transgenic mouse models appears similar to that in AD, synaptic changes at the molecular level may not completely recapitulate the pathology in humans .
To analyze the expression levels of genes associated with memory and cognition, we obtained hippocampal tissue from AD and controls 2 to 11 hours after death. This analysis reduces variables associated with cultured neuronal cells and the differences observed in animal models compared to human AD. Our results show that in AD and controls, the reduction in intracellular levels of BDNF, TrkB, DYN, and COX2 does not correlate with intraneuronal Aβ and P-tau. There may be an important indirect effect of Aβ and P-tau that reduces the levels of the expression of these genes. An alternative proposal is that Aβ and P-tau are not the primary cause of AD but that increased oxidative stress caused by other pathogenic factors may lead to the increases in Aβ and P-tau, and possibly decrease the expression of neural genes [16, 52].
This project was initiated while R.K.F. was in Dr. Paul Shapshak's laboratory (Department of Psychiatry, University of Miami Miller School of Medicine). We thank Dr. Carol Petito (Department of Pathology, University of Miami Miller School of Medicine) and Brigitte Shaw (Imaging Laboratory, Diabetes Research Institute, University of Miami Miller School of Medicine) for advice and guidance with LCM. We thank David Vasquez and Blanca Rodriguez for excellent technical assistance and Virginia Roos for editorial assistance. This work was supported in part by the South Florida Veterans Affairs Foundation for Research and Education and by the Department of Veterans Affairs. G.A.H. is the recipient of a VA Senior Research Career Scientist award.