|Home | About | Journals | Submit | Contact Us | Français|
The reasons for the selective vulnerability of distinct neuronal populations in neurodegenerative disorders are unknown. The cholinergic neurons of the basal forebrain are vulnerable to pathology and loss early in Alzheimer’s disease and in a number of other neurodegenerative disorders of the elderly. In the primate, including man, these neurons are rich in the calcium buffer calbindin-D28K. Here we confirm that these neurons undergo a substantial loss of calbindin in the course of normal aging and report a further loss of calbindin in Alzheimer’s disease both at the level of RNA and protein. Significantly, cholinergic neurons that had lost their calbindin in the course of normal aging were those that selectively degenerated in Alzheimer’s disease. Furthermore, calbindin containing neurons were virtually resistant to the process of tangle formation, a hallmark of the disease. We conclude that the loss of calcium buffering capacity in these neurons and the resultant pathological increase in intracellular calcium are permissive to tangle formation and degeneration.
A major question remains unanswered in relation to neurodegenerative disorders: what are the causes of selective vulnerability of certain neuronal populations? Here we provide evidence that age-related loss of the calcium binding protein calbindin-D28K (CB), which can lead to destabilization of calcium levels and abnormalities in calcium signaling, is a major factor allowing selective vulnerability of one such population, the basal forebrain cholinergic neurons (BFCN).
The human BFCN provide the entire cortical mantle with cholinergic innervation [30, 31] and participate in the cognitive processing of memory and attention [13, 41], which show deficits in Alzheimer’s disease (AD). The BFCN are vulnerable in many neurodegenerative disorders of the elderly, including AD, Parkinson’s disease and Lewy-body dementia [1, 2, 15, 36, 40]. In AD, the BFCN and their axons are among the earliest degenerating neural elements . The BFCN are vulnerable to phosphorylated tau accumulation and tangle formation early in mild cognitive impairment (MCI) and AD . We found that phosphorylated tau accumulation in pre-tangles, tangle formation and axonal abnormalities characterize the BFCN early in the course of aging and AD. Pre-tangles were detected in the BFCN as early as the third decade of life. A low density of pre-tangles and tangles was present in the BFCN of normal young individuals below 65 years of age. The density of tangles and pre-tangles in the BFCN displayed an increase in the brains of normal old individuals (above 65 years), and a progressive and significant increase in brains of pathologically mild and pathologically severe AD cases. In AD there is a strong correlation between loss of BFCN and severity of dementia [25, 38] and the cholinergic system is among a handful of targets for available therapy . Thus, understanding the causes of the vulnerability of BFCN is of interest both scientifically and therapeutically.
The BFCN display a number of changes, including loss of RNA and protein for high affinity nerve growth factor receptor (TrkA) , which may contribute to their degeneration. However, these changes occur concurrent with appearance of cognitive deficits in MCI and AD. Since age is the primary risk factor in AD and other neurodegenerative disorders of elderly, we speculated that age-related changes must make a significant contribution to vulnerability of the BFCN. Consistent with this expectation, we observed selective loss of the calcium binding protein calbindin-D28K (CB) from BFCN in the course of normal aging [14, 46]. The number of CB-positive BFCN and the percentage of the ChAT-positive BFCN that contained calbindin immunoreactivity were significantly lower in brains from normal individuals above 65 years of age when compared with normal individuals below 65 years. Age-related loss of CB is a primate specific phenomenon; the rodent BFCN are devoid of CB  and age-related loss of CB from BFCN also occurs in non-human primates .
Calbindin binds calcium with high affinity, regulates its intracellular levels  and protects neurons from degeneration induced by elevations of intracellular calcium [10, 37, 39]. Therefore, we hypothesized that age-related loss of CB may lead to vulnerability of the BFCN to calcium insults in neurodegenerative disorders. Here we confirm loss of CB protein from the BFCN in the course of normal aging and demonstrate that CB displays a further loss at the level of protein and RNA in AD, that loss of BFCN in AD occurs in neurons that have lost their CB in the course of normal aging and AD, and that presence of CB is associated with substantial protection against phosphorylated tau accumulation and tangle formation.
De-identified brains from 12 normal young individuals (20–64 years, 7 M, 5 F), 25 normal, non-demented old individuals (68–99, 12 M, 13 F), without any signs of neurological or psychiatric disorders, and 28 clinically and pathologically confirmed AD patients (61–99 years, 16 M, 12 F) were obtained at autopsy and used in these experiments. Detailed clinical records were available for every subject. Clinical and neuropathological diagnoses were according to the Consortium to Establish a Registry for Alzheimer’s Disease (CERAD) [33, 34]. Postmortem intervals in the three groups of subjects (young group 14.75 ± 3.4, old group 14.32 ± 1.5 and AD group 17.1 ± 1.7) and age of the old and AD groups (old group 82.7 ± 1.6 and AD group 79.2 ± 2.1) were not significantly different (p>0.05).
Series of sections from blocks containing the basal forebrain fixed in 4% paraformaldehyde for 36 hours and cut at a thickness of 40 μm were processed immunohistochemically using the avidin-biotin-peroxidase (ABC) method as described elsewhere , utilizing the Vectastain Elite ABC kit (Vector Laboratories, Burlingame, CA). The antibodies used were a specific polyclonal antibody against ChAT (gift of Dr. Louis Hersh, University of Kentucky Medical School, 1/300–1/500), mouse monoclonal and rabbit polyclonal antibodies against CB (Swant, Switzerland, 1/1000), a monoclonal antibody against p75NTR (Lab Vision Inc., Fremont, CA), and the PHF-1 monoclonal antibody that recognizes tau phosphorylated at Ser396/404 (gift of Dr. Peter Davies, Albert Einstein School of Medicine). To control for non-specific staining, sections stained using the above antibodies were compared with sections stained in the absence of primary antibodies or in the presence of non-specific IgG in place of primary antibodies.
For concurrent visualization of two antigens within the same section, the double-immunohistochemical method of Levy et al.  was used. Tissue sections were first processed for one antigen using diaminobenzidine as chromogen. After the development of the DAB brown reaction product, tissue sections were processed for the second antigen with the peroxidase labeling visualized using benzidine dihydrochloride, which results in a granular blue reaction product. Alternatively, fluorescence double labeling was performed using secondary antibodies conjugated to Texas Red and FITC. The two antibodies used in each double immunohistochemical experiment were obtained from different species (i.e. mouse and rabbit).
Matching adjacent sections single or double stained for the antigens under study were available in all immunostained cases. For this analysis, all magnocellular immunoreactive neurons in 1–2 sections spanning each of the anterior, intermediate and posterior sectors of the nucleus basalis of Meynert —cholinergic cell group 4 (nbM-Ch4), the primary component of BFCN, were counted. Counting was carried out at 10X magnification using a counting box placed in the eyepiece of a compound microscope. Data were compiled as the number of immunoreactive profiles per section of the entire Ch4-nbM.
In six cases per group, enough sections with systematic random sampling of basal forebrain stained for ChAT and CB were available satisfying the requirements for unbiased stereological estimation of immunoreactive profiles. Stereological analysis using these sections was carried out according to procedures previously described in detail , employing the fractionator method and the Stereologer (Systems Planning) software. Using this method, stereological estimates of the total number of immunoreactive profiles in each case were obtained.
Brain slabs containing the basal forebrain were prepared from fresh tissue immediately following autopsy, frozen in isopenthane cooled on dry ice, and stored at −80° C until used. Ten μm thick sections through fresh frozen basal forebrains prepared on a cryostat were stained with histogene and magnocellular BFCN were collected on caps using a laser capture microdissector. Cells were extracted from caps and rapidly dissolved in Trizol. The concentration of total RNA was measured by UV spectrophotometry, and RNA quality was confirmed in an Agilent bioanalyzer. Quantitative RT-PCR was used to investigate differential expression of the CB gene in old and AD BFCN. Two μg of total RNA from each basal forebrain was reverse transcribed to synthesize cDNA using superscript reverse transcriptase (Invitrogen). The following primers were used: CB forward GACGGCAATGGATACATAGAT, CB reverse ACTGGCCTAAGCATAGACTTTC, GAPDH forward AGGTGAAGGTCGGAGTCAACGG, and GAPDH reverse CGGTGCCATGGAATTTGCC. Amplification of each gene was expressed as the amplification cycle at which its PCR product was first detected (threshold cycle, CT). All samples were processed in parallel. The relative quantification of target gene expression was performed according to the comparative CT method. For each run, the mean expression level of the CB gene was normalized to the mean expression levels of the GAPDH reference gene. Serial dilution of standard human cDNA was used to correct for differences in amplification efficiency.
The basal forebrain area in which BFCN are located was dissected from fresh tissue immediately after autopsy, frozen on dry ice and stored at −80° C until used. Analysis was carried out on whole cell protein extracts separated on 12% SDS-PAGE and transferred to PVDF membranes. Membranes were probed with antibodies to CB and GAPDH and immunoreactivity was visualized with peroxidase conjugated secondary antibody and electrochemiluminescent (ECL) detection kit (Amersham). Bands visualized on exposed film at the correct molecular weights were quantified for protein content using the Image J software.
Data for each portion of the study were found to be normally distributed; therefore parametric statistical analyses were used. Analysis of variance followed by Bonferroni post hoc comparisons or t-tests were used to detect significant group differences. Significant relationships were analyzed using Pearson correlations.
Using immunohistochemistry for the specific cholinergic enzyme, choline acetyltransferase (ChAT, Fig. 1a and b), the low affinity neurotrophin receptor (p75NTR) and CB (Fig. 1d and e), and counts in matching sections at three levels of nbM-Ch4, we investigated the expression of these proteins in aging and AD. Both ChAT (Fig. 1a and b) and p75NTR immunoreactivity visualized a large population of the BFCN. While p75NTR was present in a slightly lower number of BFCN, counts of the two markers did not differ significantly (Fig. 1g). Additionally, counts of ChAT and p75NTR were not different in young brains when compared with the old group (Fig. 1a, b and g). In contrast, the numbers of CB-positive BFCN displayed a significant decrease when old subjects were compared with the young (Fig. 1d, e and g, Fig. 2ac). In young brains, nearly 70% of the ChAT-positive BFCN contained CB. In old brains, this percentage was reduced to 36, significantly lower than the young (Fig. 1h). Thus, we confirmed the significant age-related loss of CB from BFCN in the human brain.
There was no significant correlation between age and number/percentage of CB-positive BFCN in the young or the old group of subjects. However, when data for the two groups were combined, a significant correlation was detected between the age of the subjects and the number of CB-positive BFCN as well as the percentage of BFCN that contained CB immunoreactivity (p<0.0001 for both). Thus, it appears that gradual loss of CB occurs throughout aging and that the rate of this loss is accelerated at approximately 65 years of age (the cutoff used to separate young from old in this study), resulting in significantly lower numbers of CB-positive BFCN in the old group when compared with the young.
Consistent with the well-known loss of BFCN in AD, the numbers of ChAT and p75NTR immunoreactive BFCN displayed a significant decrease in AD brains when compared with old individuals (Fig. 1b, c and g). In contrast, the numbers of CB-positive BFCN displayed only a small and non-significant decrease in AD brains when compared with the old group (Fig. 1e, f and g, Fig. 2b–f), indicating that the neurons with preserved CB immunoreactivity during aging survive in AD. The percentage of the ChAT- and p75NTR-positive BFCN remaining in the AD brains that contained CB immunoreactivity displayed a significant increase when compared with the normal aged group and was the same as the percentage in young brains (Fig. 1h). Therefore, the loss of BFCN in AD occurs primarily in neurons that lose their CB in the course of normal aging. Importantly, the number of CB-positive BFCN did not display a correlation with PMI in the three groups of brains (p>0.05).
We next used unbiased stereological analysis in five young, five old and six AD brains that had sufficient tissue to satisfy the requirements of stereology, to confirm the pattern of CB expression using counts of the entire BFCN population (Table 1). Stereological counts confirmed the preservation of ChAT-positive BFCN in aged brains when compared with the young (Fig. 2g). In contrast, there was a significant decrease in the number of BFCN that contained CB immunoreactivity in the old brains when compared with the young group (Fig. 2a–c and g). In AD, the numbers of CB-positive BFCN remained stable despite a significant loss of ChAT-positive BFCN (Fig. 2b–f and g). The percentage of the ChAT-positive BFCN that displayed CB was significantly lower in the old group (27%) when compared with the young (71%). In contrast, this percentage was significantly higher in the AD brains (74%) and was the same as that in the young (Fig. 2h). These results confirm that the BFCN which contain CB immunoreactivity are preserved in AD.
Some modes of neuronal damage may be associated with upregulation of CB [9, 27]. To determine whether the presence of CB in nearly all AD BFCN is due to upregulation of this protein rather than preservation of CB containing neurons, we collected AD BFCN from fresh frozen tissue sections using laser capture microdissection and determined the levels of CB RNA by quantitative real-time PCR. We also determined the levels of CB protein in dissected basal forebrains from six young, old and AD brains each, using Western blot analysis. CB RNA levels were significantly lower in AD BFCN when compared with the old group (Fig. 3a). The levels of CB protein were significantly lower in the aged basal forebrain when compared with the young and showed a further significant decrease in AD (Fig. 3b and c). Thus, both CB RNA and protein display significant loss in AD BFCN, indicating no damage-induced upregulation. Despite this decrease, immunohistochemically demonstrable presence of CB identifies BFCN that survive the neurodegenerative insult.
We and others have shown that accumulation of phosphorylated tau and formation of pre-tangles and tangles in the BFCN commence very early in the course of aging, show a progressive increase in the young-old-AD continuum and are strongly related to the loss of BFCN [16, 29, 38]. We have demonstrated phosphorylated tau-positive pre-tangles within the BFCN as early as the third decade of life. The numbers of pre-tangles and tangles within the BFCN increased in old individuals (older than 65 years) and showed further progressive increase in pathologically mild and severe AD . Therefore, we sought to determine whether CB-positive BFCN in aging and AD are protected from the process of tangle formation. Counts of BFCN positive for the PHF-1 epitope of abnormally phosphorylated tau and CB in double-stained sections revealed the formation of tangles and pre-tangles in old brains (n=4, Table 2, Fig. 4 and and5).5). While the CB-positive BFCN comprised about 27% of the total BFCN population in old brains (Fig. 2), on average less than 2% of the CB-positive BFCN contained PHF-1-positive tangles and pre-tangles (Table 2, Fig. 4a–d and g–h) and only 7.5% of the total population of PHF-1-positive BFCN pre-tangles and tangles were found in CB-positive neurons. Thioflavin S, which binds the abnormal β-pleated sheet protein conformations and is indicative of the formation of mature tangles, displayed a similar pattern. Less than 1% of the CB-positive BFCN contained thioflavin S-positive tangles in old brains and only 3.7% of thioflavin S-positive tangles within the aged BFCN were in CB-positive neurons (Table 2, Fig. 5a). While slightly higher, counts of PHF-1 immunoreactive pre-tangles and tangles and thioflavin S-positive tangles in AD BFCN were similar to that in the aged group. Even though CB-positive neurons comprise the overwhelming majority of the BFCN in AD (Fig. 2), on average only 5.2% of CB-positive BFCN contained PHF-1-positive tangles or pre-tangles in AD brains and only 1.9% of the PHF-1-positive BFCN tangles and pre tangles were in CB-positive neurons (Table 2, Fig. 4c, e and i–k, Fig. 5c and d). Similarly, on average 2.4% of CB-positive BFCN contained thioflavin-S-positive tangles and only 2.8% of thioflavin-S-positive tangles in BFCN were in CB-positive neurons. Therefore, CB-positive BFCN appear to be relatively protected from the process of tangle formation in aging and AD.
The results of the present set of experiments clearly demonstrate that the presence of CB within the human BFCN confers protection against degeneration and tangle formation in aging and AD, and that age-related loss of calbindin from the BFCN identifies neurons destined to degenerate in AD. While the number of BFCN displayed a significant decrease in AD brains, the numbers of CB-positive BFCN remained the same when compared with the normal aged group, resulting in increased percentage of the remaining ChAT-positive BFCN that contained CB. Furthermore, a very small percentage of the total numbers of BFCN tangles and pre-tangles were detected in CB-positive BFCN in aging and AD. CB RNA and protein showed a significant decrease not only in the aged but also in AD BFCN, indicating that the presence of CB within the AD BFCN is not due to upregulation of this protein induced by neurodegenerative damage, but rather due to preservation of CB containing neurons.
In addition to binding calcium with high affinity, CB serves as a calcium sensor and upon binding calcium, interacts with a number of other proteins [23, 32]. The presence of intense CB immunoreactivity in the primate BFCN indicates the dependence of these neurons on the calcium buffering and calcium sensing functions of this protein. Of great interest, while the age-related loss of CB from the BFCN may potentially interfere with the normal functions of these neurons, it does not automatically result in their degeneration since we found the number of BFCN to remain stable in the aging process. Neurodegenerative insults seem to be required to cause degeneration of the BFCN that have lost their CB in the course of aging. In particular, it is noteworthy that the amyloid-β peptide (Aβ), which accumulates in aged and AD brains, results in significant dysregulation of intracellular calcium levels [12, 28]. Thus, the age-related loss of CB from the BFCN appears to deprive these neurons of a major mechanism of intracellular calcium buffering and signaling. This leaves the BFCN vulnerable to the calcium dysregulation that follows neurodegenerative insults, such as that caused by Aβ accumulation, and results in the selective vulnerability of these neurons to degeneration.
The human BFCN are exquisitely vulnerable to formation of pre-tangles and tangles in aging and AD [16, 29]. In this process, abnormal phosphorylation of tau leads to its aggregation and its dissociation from microtubules [5, 42]. Aggregation of abnormally phosphorylated tau in pre-tangles and tangles, microtubule disintegration and potential direct toxic effects of tau oligomers  result in neuronal dysfunction and degeneration. Our findings and those of others [4, 19] indicate that calcium dysregulation is permissive of this process. They also suggest that restoration of calcium homeostasis is likely to protect the BFCN from loss in neurodegenerative disorders. CB interacts with the plasma membrane calcium pump . Thus, therapeutic strategies that target calcium pumps may potentially protect neurons from calcium dysregulation and degeneration.
The age-related loss of CB from the BFCN is a neurochemically specific phenomenon. The BFCN in old brains displayed preserved ChAT and p75NTR immunoreactivity when compared with the young. This loss also appears to be regionally specific. Substantial age-related loss of neuronal CB has not been reported elsewhere in the nervous system. Our own relatively exhaustive survey of CB-positive neurons of the human cerebral cortex revealed preservation of these neurons in 13 of 17 cortical areas investigated, and statistically significant loss of CB-positive neurons ranging between 20–46% in only 4 cortical areas (visual association cortex, posterior cingulate cortex, primary visual cortex and parahippocampal gyrus) . Similar age-related loss of CB from the BFCN in non-human primates  indicates that this phenomenon is related to the normal aging process and not a consequence of pathological processes or injury.
The BFCN are not the only neuronal population vulnerable to neurodegenerative insults. While the vulnerability of the BFCN in several neurodegenerative disorders of the aged [1, 2, 15, 36, 40] indicates that the pathological insults causing these disorders may all be conducive to calcium dysregulation, it remains to be seen whether this mechanism is operative in all vulnerable neurons. In AD, neurons in the entorhinal cortex are similar to the BFCN in that they are vulnerable to pathology and loss early in the course of the disease. They appear to be the first neurons within which pre-tangles and tangles are formed  and they are lost in very early AD , before frank signs of dementia become severe. There is evidence that the entorhinal neurons may also suffer from calcium dysregulation in AD. Loss of GluR2 subunit of AMPA receptors has been observed in neurons of layer 2 of entorhinal cortex in AD , prior to formation of tangles or neuronal loss. GluR2 is a calcium impermeable AMPA receptor subunit and its loss will result in activation of the calcium permeable GluR1 and 4/5 subunits, thus potentially leading to increased intracellular calcium prior to tangle formation and cell death. Of interest, age-related loss of GluR2 from the BFCN has also been observed in the human brain . Elucidation of further potentially shared calcium dysregulation mechanisms among various vulnerable neuronal populations in neurodegenerative disorders must await further investigations.
While age-related calcium dysregulation was hypothesized as the basis of neurodegenerative disorders over a decade ago , our observation of the loss of CB is the first clear calcium-related pathway identified that is associated with selective vulnerability of one neuronal population in these disorders.
We are grateful to Girgis Girgis and Katherine Gasho for expert technical assistance. This work was supported in part by a Zenith Fellows Award (C.G.) from the Alzheimer’s Association; and by grants from the National Institute on Aging (AG014706 and AG027141). A portion of the tissue used in these studies was received from the Northwestern University (AG013854) and Massachusetts General Hospital (AG005134) Alzheimer’s Disease Centers.
Conflict of Interest
The authors declare that they have no conflict of interest.