| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
From the Cellular and Molecular Neuropathology Program,* Center for Neurobiology and Behavior, Department of Psychiatry, University of Pennsylvania, Philadelphia, Pennsylvania; the Computer Vision Laboratory for Vertebrate Brain Mapping, Department of Neurobiology and Anatomy, Drexel University, Philadelphia, Pennsylvania; the Rush Alzheimer’s Disease Center and Department of Neurological Sciences, Rush-Presbyterian St. Luke’s Medical Center, Chicago, Illinois; and the Center for Neurodegenerative Disease Research and Institute on Aging, University of Pennsylvania, Philadelphia, Pennsylvania
 |
Abstract |
|---|
 Top Abstract Materials and MethodsResultsDiscussionReferences |
|---|
Dendrites, in particular, are dynamic structures critical for the fidelity of neural transmission. They undergo substantial modifications in morphology and molecular composition in response to afferent neural transmission, trophic stimulation, or pathological processes.5 The extent to which degeneration and loss of dendritic processes occur in AD is not well understood, nor is the relationship of dendritic changes to PHFtau, Aß, neuron loss, or the selective vulnerability of particular regions for degeneration in AD. This is at least partially due to the methodological challenges of studying dendritic architecture. Golgi studies of small numbers of patients have reported loss of dendritic extent in the subiculum, CA-1, and dentate gyrus in AD.1 However, the classical Golgi method stains relatively few neurons and while the morphological detail can be excellent, the approach does not lend itself to larger scale analyses or correlation of dendritic indices with other markers labeled in the same or adjacent sections.
The region-specific vulnerability to degeneration in AD is another aspect of the disease that is still poorly understood and for which a study of dendritic arborization may shed light. Neurofibrillary changes and accompanying neuron loss in AD exhibit a selectivity for particular brain regions that follows a predictable sequence which mirrors the connectional anatomy of affected brain regions and the severity of clinical symptoms.6,7 The earliest and most severe changes occur within the perirhinal and entorhinal cortices, followed by the hippocampus (especially CA1) and amygdala, then other limbic and neocortical association areas, and finally primary motor and sensory cortices. Intrinsic selectivity is also evident in terms of specific lamina and subregions being more affected than others. However, the relationships between degenerative changes in various neural elements (eg, dendrites, axons, neuronal somata) within a given region have not been well characterized.
The subiculum may be a particularly instructive region of interest because of its relatively simple and well-studied cytoarchitectural and connectional neuroanatomy and also its selective vulnerability for AD-related neurodegeneration. Although mildly affected in early stages, the subiculum is severely affected by the pathological changes that correspond to the advanced stages of AD (Braak stages IV to VI), with marked deterioration during the progression to stages V to VI.6,8 As a major source of hippocampal efferents to widespread cortical and subcortical structures, the subiculum acts as a final station in the sequence of projections that link the subdivisions of the hippocampus and also receives direct projection from the entorhinal cortex (EC).9 This important role in the ventromedial temporal lobe memory system implies that damage to the subiculum could have profound functional consequences and explains the correlation of subicular damage with severity of AD symptoms.10
In this study, an antibody against MAP2, a robustly expressed cytoskeletal protein that is localized to dendrites, was used to immunolabel dendrites with the goal of examining the extent of MAP2-labeled dendritic arborization in relation to neurodegenerative pathology in connectionally relevant anatomical subregions of the subiculum. Dendritic changes in the subiculum in AD were quantified with a new "dendritic arborization index" in which we used random, systematic sampling for image capture and customized image analysis to determine the percent area occupied by dendrites. We also quantified the density of neurons, NFTs, and Aß deposits to assess their relationships to diagnosis and dendritic arborization.
|
Materials and Methods |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
Brain specimens from 15 subjects with AD and 15 comparison subjects with no neuropsychiatric disease were obtained at autopsy from the Alzheimer Disease Center Core of the University of Pennsylvania. All AD subjects were in a moderately severe to end-stage of dementia at the time of death. Clinical information is provided in Table 1
.11,12
All hippocampal sections were taken from comparable levels of the uncal region of the hippocampus and were confirmed by comparing Nissl stained sections with a standard atlas.13
Brain sections were taken from levels 1 to 7 in Duvernoy’s parcellation. There were no significant differences between the groups for age, sex, postmortem interval, or brain weight. Gross and microscopic neuropathological examinations were conducted for all cases and the diagnosis of AD was made using established diagnostic criteria.14
The brains of four normal subjects had minor neuropathological findings that were not relevant to this study (two with resolved lacunar infarcts that did not involve the hippocampus, one with cerebellar neuron loss, and one with a subdural hematoma). The brains of the AD patients were without neuropathological abnormalities aside from abundant senile plaques and NFTs.
|
Coronal blocks of the temporal cortex were dissected at autopsy and fixed in ethanol (70% EtOH, sodium chloride 150 mmol/L) for 24 hours. The blocks were then paraffin-embedded and cut into 10-µm-thick sections for serial staining and analyses. For neuronal density measurements, tissue was immunolabeled with NeuN (Chemicon, Temecula, CA), a monoclonal antibody specific for neurons. MAP2-immunoreactive neurons and dendrites were labeled with M13, a specific mouse monoclonal antibody that identifies all isoforms of MAP2 (generously provided by V. M.-Y. Lee). Neurofibrillary tangles were labeled with a phosphorylated tau-specific monoclonal antibody, AT8, which recognizes PHFtau specifically containing a phosphorylated Ser202 residue (InnoGenex, San Ramon, CA) and Aß plaque deposits were labeled with anti-Aß, a monoclonal antibody raised against a synthetic peptide of residues 8 to 17 in the amyloid ß-protein/amyloid A4 present senile plaques of all types. (DAKO, Carpinteria, CA). Immunohistochemistry for MAP2 was performed according to a previously described procedure15 using the purple VIP reporter (Vector Laboratories, Burlingame, CA) which gives excellent resolution for quantitation of dendrites. Sections from all cases were run simultaneously for each antibody on an automated immunohistochemical stainer (BioGenex, San Ramon, CA) in precisely timed runs. Antigen retrieval pretreatment included 90% formic acid for 3 minutes before anti-Aß and for 5 minutes before AT8 incubation. No pretreaments were used for NeuN or M13.
Apportionment of the Subiculum
To assess possible differences in dendritic arborization based on connectional patterns within the subiculum, individual measurements were taken in relevant subportions of the subiculum. Cytoarchitecturally, the subiculum is a simple, three-layered allocortical structure consisting of a molecular layer, pyramidal cell layer, and fusiform cell layer (Figures 1 and 2)
. Each layer is easily distinguished with MAP2 staining and has its own connectional anatomy. For example, in non-human primates, afferents from the EC synapse primarily in the superficial part of the molecular layer and the fusiform layer of the subiculum, while CA1 afferents synapse throughout the molecular and pyramidal layers.9,16
Furthermore, projections from CA1 and the EC to subiculum in the monkey are organized along the transverse axis of the subiculum. Rostral and lateral EC projects to the proximal subiculum (close to CA1), while more caudal and medial portions of the EC project to the distal subiculum (close to presubiculum).17,18
For both the proximal and distal halves of the subiculum, measurements were made for the superficial and deep portions of the molecular layer, the pyramidal cell layer, and the fusiform cell layer.
|
Estimation of Neuron and NFT Densities
Quantitation of neurons and NFTs within the pyramidal and fusiform layers was performed with a stereological mapping station which included a Leica DMRBE microscope and computer equipped with StereoInvestigator software version 3.23 (MicroBrightField, Inc., Williston, VT). After cytoarchitecturally delimiting the region of interest at x200 magnification, a grid of predetermined size was placed over the entire region. Magnification was increased to x400 and the program stopped at each intersection point on the grid for counting. The operator focused through the section depth as the fields were visualized on the video monitor within the superimposed counting frame. All objects within the 150 x 150 µm counting frame that did not touch the exclusion lines of the box (bottom and left sides) were counted. Approximately 50% of the area of a section was quantified.
For neuron density estimates, neurons were identified based on NeuN labeling. Neurons were counted only if the nucleus was visible in the section to reduce split-cell artifact due to variable neuronal profile size. NFTs were labeled with AT8 and had a characteristic "flared" appearance due to their location in the cell bodies of pyramidal neurons or pyramidal ghost tangles.
Quantitation of Dendritic Arborization and Aß Plaque Burden
Using the random systematic sampling scheme, images were captured from approximately 40% of a defined region of interest for estimation of MAP2 dendritic arborization within the fusiform and pyramidal cellular layers, as well as the deep and superficial portions of the molecular layer. As for the neuron and NFT counts, separate measures were made in the proximal and distal halves of the subiculum. For Aß deposits, subparcellation was not performed. A custom analysis algorithm provided a value for the area occupied by the Aß-immunopositive zone or the area fraction, as previously described.19,20 This is defined as the mean percent fraction of labeled pixels per field. The sensitivity of this measure to illumination intensity was addressed by adhering to a fixed protocol of light calibration. Briefly, 24-bit images were captured at each sampling point and converted to 8-bit images by using Adobe Photoshop 5.02 (Adobe Systems Inc., San Jose, CA). Area analysis was performed using macros designed for the public domain Object-Image 2.06 (Norbert Vischer, http://simon.bio.uva.nl/object-image.html). Overall data reduction yielded a value for the area fraction defined as the percent fraction of Aß-immunolabeled deposits to total tissue area of the defined region of interest. The MAP2 dendritic arborization area was then expressed as a ratio of area per NeuN neuron density. This represents an index of dendrites per neuron.
Laminar Thickness and Correction for Atrophy
The thicknesses of the molecular (apical dendritic), pyramidal, and fusiform cell layers were measured using StereoInvestigator software. The thickness of each layer was measured at the proximal (close to CA1) and distal (close to presubiculum) ends of the subiculum.
Atrophy of the hippocampus occurs with AD and this would artificially increase the densities of dendrites, neurons, NFT counts, and Aß plaque burden. To minimize this confound, raw density indices were corrected based on laminar thickness measurements for each case. Ideally, we would correct for the three-dimensional volume of the hippocampus if this were known. However, the manner in which our tissue was processed did not allow accurate measurement in the transverse or longitudinal planes because of varying methods of block dissection at the autopsy table. Therefore, a correction factor for each case was generated based on our coronal plane thickness measurements. The raw density index for each marker was multiplied by the correction factor, L x T, where L = the thickness of the layer being measured and T = the average total thickness of the subiculum (from pial surface to gray/white matter border). This two-dimensional correction is a conservative estimate for the three-dimensional atrophy that would contribute to the raw density measurements.
Data Analysis
Between group differences for demographic variables were assessed with Student’s t-tests while differences for MAP2 dendritic arborization indices, laminar thicknesses, and densities of NeuN-stained neurons, NFTs, and Aß plaque deposits were analyzed with Wilcoxon sign-rank tests. Non-parametric statistics were used because of small sample size and to protect against outliers. Relationships between neuroanatomic markers were assessed with Spearman rho (
) correlations. For the initial between-group analyses, data were reduced into mean dendritic arborization indices for the molecular (deep and superficial), pyramidal, and fusiform layers using an average of the proximal and distal measures. Secondary analyses then examined differences within the individual subregions as well as differences between deep and superficial portions of the molecular layer in each subregion.
|
Results |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
The subiculum of AD cases showed a significant reduction in the average total thickness by approximately 33% (Table 2)
. However, there were differences between layers where the molecular layer was reduced 43% while the pyramidal cell and fusiform layers were reduced only 25% and 23%, respectively. Given the largely synaptic composition of the molecular layer compared to the cellular layers, this disproportionate reduction would be consistent with the hypothesized dendritic loss in AD. The proximal portion of the subiculum exhibited slightly greater reduction in laminar thickness for its molecular layer compared to the distal molecular layer (51% that of control in the proximal subiculum versus 65% in the distal subiculum) while the opposite was true for the pyramidal cell layer (82% proximal versus 65% distal). Fusiform atrophy of 22% was similar in proximal and distal subiculum.
|
Raw dendritic arborization indices were similar between groups in all three layers: molecular layer average 7.8% (SD = 4.2%) in AD and 7.7% (7.6%) in controls; pyramidal layer 11.4% (5.1%) in AD and 12.0% (6.0%) in controls; fusiform layer 8.3% (5.3%) in AD and 7.4% (5.4%) in controls. Corrected for atrophy, significant reductions in the dendritic arborization index were observed in all layers of the subiculum as a whole, and in both proximal and distal portions (Figure 2
; Table 3
). Differences were especially robust in the molecular layer where dendrites were reduced by an average of 60% in AD compared to control cases. Furthermore, in qualitative analyses, molecular layer dendrites often appeared dystrophic. The distal portion of the subiculum exhibited slightly greater decreases in the dendritic arborization index than the proximal portion. Because the EC projections concentrate in the superficial portion of the molecular layer of the subiculum, we compared groups for the relative dendritic arborization index in the deep versus superficial portions of the molecular layer with the hypothesis that the superficial portion of the molecular layer would be more depleted of dendrites than the deep, given the well-recognized severity of entorhinal degeneration. The deep/superficial ratio for the subiculum as a whole showed no differences between groups, with a ratio of 1.4 for the AD group compared to 1.6 for the controls (Z = 1.33, P = 0.18). Similar results were observed in proximal and distal subdivisions.
|
|
The density of neurons in the subiculum (pyramidal and fusiform layers combined, Table 4
) in AD was decreased approximately 50% without noteworthy difference between proximal or distal sectors. As expected, the densities of NFTs were significantly increased in AD throughout the subiculum, approximately four-fold compared to control brains. The density of NFTs in the proximal sector of the subiculum in AD was almost twice that of the distal sector, whereas similar low levels in both sectors were seen in the control group. The Aß plaque burden was similarly increased in AD.
|
We assessed the relationship of dendritic arborization to NFT counts and Aß plaque burden in the subiculum for the whole sample. The average dendritic arborization index (mean of proximal and distal values) negatively correlated with NFT density in the pyramidal layer (Spearman’s = -0.59, P = 0.0009) and fusiform layer ( = -0.46, P = 0.015) and approached significance in the molecular layer ( = -0.36, P = 0.06). Subregional analyses found the correlations to be most prominent in the distal half of the subiculum. There were no significant correlations between dendritic arborization index values and Aß plaque density values.
NeuN-immunoreactive neuron density also negatively correlated with NFT density in the subiculum as a whole ( = -0.70, P < 0.0001) and for proximal ( = -0.71, P < 0.0001) and distal halves ( = -0.68, P < 0.0001). There was a marginally significant decrease in neuron density in relation to Aß plaque density ( = -0.47, P = 0.05).
|
Discussion |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
In this study of AD and elderly control subjects, raw dendrite densities obtained by the algorithm did not differ between the Alzheimer and control group. However, marked atrophy in AD was visually apparent and quantitatively resolved in the large reductions in laminar thicknesses. The volume of a given region or layer is a function of the space-occupying elements of which it is composed, including neuronal and non-neuronal cells and their associated processes, as well as extracellular matrix. Space occupied by dendrites exceeds that of other neuronal or non-neuronal elements within the cortex. This is especially true in the molecular layer which is predominantly dendritic. Our data indicates that the reduction in laminar thickness, which was most dramatic in the molecular layer, is predominantly due to the loss of dendrites. If it were due to loss of other elements or artifactual shrinkage (eg, by differences in dehydration between groups during processing), then the raw dendritic arborization indices would be increased. These considerations indicate the necessity of correcting the raw measures of dendrite material for estimates of normalized volume, or some index related to volume, rather than raw densities obtained by image analysis of individual sections alone. Our data show significant reductions in the amount of dendrites in the subiculum in AD.
Stereologic studies of the subiculum have shown reductions in neuronal number in AD, particularly in later stages.8,22 Our available material did not allow us to determine the total volume of the subiculum and so we could not estimate the total numbers of neurons or NFTs (or total loads of Aß or MAP2 dendrites) in the subiculum. However, our data do provide support for marked neuron loss in the subiculum in AD in that the density of neurons corrected for atrophy was significantly reduced.
As expected, we found that the density of NFTs was significantly higher in AD than control subiculum. Of particular note is that dendritic arborization indices negatively correlated with the densities of NFTs in the subiculum as a whole, and in the distal sector in particular. In contrast, while the Aß plaque burden was also increased in AD, we observed no significant correlations with dendritic arborization. This suggests that neurofibrillary degeneration is more directly related to dendritic pathology than is Aß. Negative correlations of neuron density with NFTs and Aß plaque burden were also observed.
Dendritic loss in AD may be due to a number of causes. First, as supported by our correlation data, it is probable that accumulation of PHFtau in the form of NFTs and neuropil threads in dendrites leads to retraction and loss of dendritic arbors. Another possibility is a physical or toxic effect of Aß on dendrites, either in its insoluble or soluble forms. Tau and MAP2-immunoreactive neuropil threads, present in dendrites (and axons) are a prominent component of neuritic plaques. Adlard and Vickers2 found an increased MAP2 immunodensity in diffuse plaques and normal or decreased MAP2 in fibrillar or cored plaques. Neuron culture experiments with Aß indicate a complex effect on neurites, with both trophic and dystrophic effects.3,23 In our study, Aß plaques of all types were sparse in the AD subiculum, occupying less than 1% of the area, and there was no correlation with MAP2 dendrite values. Thus, it is unlikely that insoluble Aß plaques are directly related to dendritic loss, although an effect of soluble Aß cannot be excluded. A third possibility is that loss of dendrites may occur due to deafferentation.24 Experimental disruption of the EC projection to the dentate gyrus has been shown to cause a reduction in the extent of deafferented granule cell dendrites.25 The extent of these dendrites is also reduced in AD,24 suggesting that dendrite degeneration secondary to deafferentation may occur in AD. Furthermore, phosphorylated tau accumulations appear in the outer portion of the molecular layer of the dentate gyrus at Braak stage III, while NFTs and staining of the inner dendrites and mossy fibers of the granule cells does not occur until stage V. This corresponds with the termination zone of the perforant path which projects to the outer portions of the molecular layer of the dentate gyrus. These data suggest that changes in distal dendrites might precede cell body changes in dentate gyrus neurons.
The major afferent innervations of the subiculum are from CA1 and the EC (Figure 1)
. These are two of the most severely affected regions of all in AD,26 even more so than the subiculum.8
Thus, secondary dendritic retraction due to deafferentation may also contribute to the diminished dendritic arborization in the subiculum.
Topographical studies of neurofibrillary changes in AD suggest a relationship between AD progression and known patterns of anatomical connectivity.7,26 Informed by this connectivity, larger studies examining dendrite loss in the subiculum and highly linked regions across a spectrum of normal aging, mild cognitive impairment and dementia (representing sequential stages of degeneration) could test this hypothesis. These studies would help clarify the pathoanatomic sequence of synaptic, dendritic, and neuronal degeneration, PHFtau and Aß lesions. If supported, subsequent identification of the molecular mechanisms by which such transynaptic degeneration might occur offers a focus for the development of novel treatments to arrest the process in its early stages.27
|
Footnotes |
|---|
Supported by grants AG10124, MH57401, AG09215, AG15819, and AG17917 from the National Institutes of Health.
Accepted for publication June 27, 2003.
|
References |
|---|
|
TopAbstractMaterials and MethodsResultsDiscussionReferences |
|---|
This article has been cited by other articles:
 |
|
 |
|
  B. Bandyopadhyay, G. Li, H. Yin, and J. Kuret Tau Aggregation and Toxicity in a Cell Culture Model of Tauopathy J. Biol. Chem., June 1, 2007; 282(22): 16454 - 16464. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
