Dense-Core Senile Plaques in the Flemish Variant of Alzheimer’s Disease Are Vasocentric

Family 1302

The APP692 (1302) family is a multigeneration Dutch family whose members have presenile dementia and cerebral hemorrhage, inherited in an autosomal-dominant pattern (Figure 2) ▶ . 18 The clinical phenotypes overlap because hemorrhagic stroke was reported in an offspring of a demented patient, and conversely, progressive dementia has also occurred in an offspring of a stroke patient. 42 Two patients (IV-2 and IV-5) had strokes and were diagnosed with AD in their lifetimes, for instance, individual IV-2 first had dementia at age 41 years and later suffered a hemorrhagic stroke at age 46 years. 42 We earlier studied two patients who had clinical dementia and fulfilled the neuropathological criteria of AD (patients IV-4 and IV-13) (Table 1) ▶ . 18,26,43 Here, we report the neuropathological analysis of an additional member of the family, patient IV-5, who had hemorrhagic stroke at age 42 years. 18 A biopsy taken while evacuating a large hematoma in the left parieto-occipital cortex, revealed CAA and both senile and diffuse plaques, however, neurofibrillary tangles or hyperphosphorylated tau (AT8)-positive neurites were absent. Consistently, before the intracerebral hemorrhage, this patient did not show any signs of cognitive impairment. However, in the course of disease, the patient developed progressive dementia and at age 48 years was diagnosed with dementia indistinguishable from AD. 42 The patient slowly progressed to a vegetative state and died at age 55 years. The brain was fixed in 4% formaldehyde and analyzed.

Histology Including Immunohistochemistry

Examination of the brain of patient IV-5 was performed after a postmortem interval of 51/2 hours. The right cerebral hemisphere was fixed by immersion in 10% formalin and embedded in paraffin. Five-μm thick sections were taken from superior frontal gyrus, superior temporal gyrus, superior occipital gyrus, superior parietal lobule, hippocampus and entorhinal cortex, basal ganglia, midbrain with substantia nigra, pons, cerebellum, brain stem, and cervical spinal cord. Sections were examined by routine histopathological methods and also with Thioflavin S (ThS), Congo red, Cresyl violet, periodic acid-Schiff, and Bielschowsky.

For Aβ subspecies identification and other immunohistochemistry (Table 2) ▶ , serial 4- to 5-μm sections were sliced from the hippocampus, superior temporal gyrus, superior frontal gyrus, and cerebellum of the following cases: three AD/Fl patients (IV-4, IV-5, and IV-13), sporadic AD patients (n = 5), AD with PS1 mutations (I143T; n = 5 and A384A; n = 5), and a HCHWA-D patient.

For immunohistological study of the SPs in three AD/Fl patients, serial 2- to 3-μm-thick sections were sliced from paraffin-embedded blocks of the superior frontal gyrus, superior temporal gyrus, and cerebellum. On an average, 500 sections were sliced from each block. The sections were double immunostained with Aβ antibodies and blood vessel markers [CD31, CD34, smooth muscle actin or collagen type IV (C-IV)] (Table 2) ▶ . From the two neocortical and one cerebellar region, 300 and 200 SPs, respectively, from each patient (n = 3) were serially imaged by a digital charge-coupled device color camera (Sony Corporation, Tokyo, Japan) connected to a Vidas image analysis frame grabber (Kontron, München, Germany). SPs and other amyloid deposits were serially studied. SPs were so addressed if on any section had the appearance exemplified (Figures 3 and 4) ▶ ▶ , thus closely resembling such deposits in other familial and sporadic AD patients (Figure 3L) ▶ .

Antigen retrieval for Aβ immunohistochemistry was performed on sections treated in 98% formic acid for 5 minutes at room temperature. All dilutions were made in 0.1 mol/L of phosphate-buffered saline with 0.1% bovine serum albumin. Staining for single antigen was performed using streptavidin-biotin-horseradish peroxidase using chromogen 3′,3′diaminobenzidine (Roche Diagnostics, Vilvoorde, Belgium), described previously. 29 Immunohistochemistry involving detection of more than one antigen was done using species-specific or IgG subtype-specific secondary antibodies, conjugated directly to biotin, horseradish peroxidase, alkaline phosphatase, or galactosidase (Southern Biotechnology, Birmingham, AL). This was followed by color development using one of the following chromogens (acquired from Roche): 3′,3′diaminobenzidine, 3-amino-9-ethylcarbazole, Fast-red, 5-bromo-4-chloro-3-indolyl-phosphate/nitroblue tetrazolium solution, or 5-bromo-4-chloro-3-indolyl-d-galactopyranoside (X-gal). Different cycles of primary, secondary, and tertiary antibodies, and of chromogens were used during double-immunohistochemical staining to avoid bias of staining for any one particular antibody.

Using antibodies specific for Aβ N-terminus (eg, 6E10), and contrasting with reactivity for Aβ17–24 (4G8), distinguished full-length and N-truncated Aβ, as described previously. 44 Specificity of Aβ antibodies was examined by dot blotting using 50 ng of synthetic wild-type, Flemish, and Dutch Aβ peptides either as full-length Aβ (Biosource, Nivelles, Belgium) or N-truncated Aβ (12–42) peptide. Although 4G8 recognized the Flemish and Dutch Aβ less avidly compared to the wild type, no difference was observed for any N- or C-terminus Aβ antibody used in this study, or between any full-length wild type, or mutated Aβ40 or Aβ42, and their corresponding N-truncated forms (data not shown).

Fluorescence and High-Resolution Transmission Electron Microscopy

Labeling for confocal laser-scanning microscopy was done on 10- to 50-μm-thick sections incubated overnight with 4G8, or mouse Aβ40 or Aβ42 antibody, washed and labeled with an anti-mouse tetramethylrhodamine B isothiocyanate-conjugated antibody (Molecular Probes, Eugene, OR). For multiple labeling, sections were co-incubated overnight with rabbit anti-Aβ40 or anti-Aβ42 and mouse IgG₁ against vessel components (varied combinations of CD31, CD34, smooth muscle actin, and or C-IV), washed, and labeled with an anti-mouse tetramethylrhodamine B isothiocyanate conjugate and an anti-rabbit fluorescein isothiocyanate antibody. Images were acquired with a Zeiss CLM-410 (Carl Zeiss NV-SA, Zaventum, Brussels) using either 488 nm line of argon single laser or 632 nm helium-neon double laser for excitation. Three-dimensional reconstructions were made by AnalySIS (Soft Imaging System, Münster, Germany).

Araldite blocks of neocortex, hippocampus, and cerebellum of patients IV-4 and IV-13 were used for electron microscopy. Immersion fixation was achieved with 4% neutral buffered glutaraldehyde followed by 2% buffered osmium tetraoxide. Blocks were sectioned with a Reichert Jung microtome (Leica, Wein, Austria) equipped with a section counter, and ribbons of 0.25-μm-thick sections of small regions of interests were collected on copper grids. Sections were contrasted with routine uranyl acetate and lead citrate, and 20 SPs were analyzed by a Philip CM10 electron microscope (Philip, Eindhoven, The Netherlands) equipped with a goniometric coordinator. The ultra-thin sections were interspersed with thicker 1-μm sections collected on glass slides for light microscopy.

Morphometric, Densitometric, and Mass Spectrometric Analyses

Morphometric and densitometric analyses were done by a self-written software on a Vidas Image Analysis System, described previously. 45 Sizes of SPs and CAA (n = 300 each) from various regions of AD/Fl patients were measured and compared by a two-tailed unpaired t-test. Amyloid-laden vessels were ascribed severely stenotic when the ratio of the lumen diameter and the vessel diameter was less than one half.

For Aβ subspecies identification, sections were stained with different Aβ40 and Aβ42 antibodies (Table 2) ▶ . Semi-interactive quantification was done as described 45 and the percentage areas of reactivity in vessels and parenchymal plaques were assessed.

Isolation of Aβ from meningeal vessels in AD/Fl and sporadic AD brains, frozen at −70°C, was performed as described 46 with slight modification. Different parenchymal amyloid deposits were carefully extracted with tissue microdissection aided by immunohistochemistry on adjacent sections. Samples were thawed and washed three times in ice-cold Tris-buffered saline and homogenized in a buffer of 150 mmol/L NaCl, 50 mmol/L Tris-HCl, pH 8.0, containing protease inhibitors (ethylenediaminetetraacetic acid-Na, 2 mmol/L; leupeptin, 10 μmol/L; pepstatin, 1 μmol/L; phenylmethyl sulfonyl fluoride, 1 mmol/L; TLCK, 0.1 mmol/L; TPCK, 0.2 mmol/L). The homogenates were centrifuged at relative centrifugal force (RCF) 100,000 × g for 1 hour and brain tissue pellets were washed three times with ice-cold Tris-buffered saline and then extracted using 1.0 ml of 70% formic acid by sonication and vortexing for 2 hours at 4°C. The formic acid extracts were centrifuged at 100,000 × g for 2 hours and the formic acid layers were collected and stored at −20°C. From these samples, Aβ was immunoprecipitated using 4G8 and protein G Plus/Protein A agarose beads (Oncogene Science, Inc., Cambridge, MA) and analyzed using a matrix-assisted laser-desorption/ionization-time-of-flight (MALDI-TOF) mass spectrometer (Voyager-DE STR BioSpectrometry Workstation, PE/PerSeptive Biosystem) described previously. 47,48

Pathological Confirmation of AD for Patient IV-5

Weight of the brain at autopsy was 855 g and showed atrophy of all cortical areas. Old cystic infarcts were present in temporal and occipital cortices bilaterally. Microscopic examination of the right cerebral hemisphere revealed diffuse cortical atrophy and microspongiosis of upper cortical layers. Silver impregnation and Aβ immunohistochemistry revealed a huge load of CAA, SPs with large plaque cores, and diffuse plaques, in hippocampus, parahippocampus, neocortex, basal ganglia, and cerebellum. Amyloid deposits were manually counted on 4G8-stained sections by a Zeiss MC80 microscope equipped with an ocular graticule and a ×10 objective (1.56 mm²). On an average, ≥10 SPs were present in the gray matter fields of all neocortical regions as well as in hippocampus, and ≥4 SPs were present in basal ganglia, substantia nigra, and cerebellum. In other brain regions, including the central gray matter of the spinal cord, at least some forms of ThS(+) plaques were observed. AT8(+) dystrophic neurites and neuropil threads were also present in varying severity in all regions except cerebellum, where only Ubi(+) dystrophic neurites were observed. Especially in the neocortical and hippocampal regions, both SPs and CAA were equally associated with Ubi(+) and AT8(+) dystrophic neurites in the surrounding parenchyma. Silver impregnation also recognized a large number of neurofibrillary tangles in hippocampus and neocortex. Based on the presence of tangles in all sectors of hippocampus and subiculum as well as a severe involvement of isocortex, the brain was assigned Braak and Braak 49 stage V/VI (Table 1 ▶ and Figure 3 ▶ ; A to D).

AD/Fl Pathology

Core-containing SPs and coreless primitive plaques were the most abundant plaques in the neocortical and limbic areas in the three AD/Fl brains, where they were far more common in the gray, than in the white matter (Figure 3 ▶ ; E to H). SPs were associated with unusually large plaque cores and were also sometimes multicentric. However, a small number of SPs had a relatively smaller plaque core size compared to the surrounding coronal plaque and thus closely resembled SPs commonly described in AD. 2,26 SPs were also predominantly present in the granular layer of cerebellum. The primitive plaques as described elsewhere 2 were circumscribed clusters of ThS(+) amyloid wisps without a central dense core and also had a striking textural resemblance to the perivascular and coronal plaques.

A huge load of amyloid was also deposited in capillaries and small- to middle-sized arteries often displaying vessel-within-vessel configurations and frequently associated with microvascular changes such as microaneurysms, fibrinoid necrosis, and small hemorrhages (Figure 3I) ▶ . These microvascular changes have earlier been observed in HCHWA-D patients in which they correlate with the severity of the amyloid angiopathy. 50 The perivascular amyloid plaques, however, were different in the AD/Fl and HCHWA-D brains. Whereas in AD/Fl, copious perivascular amyloid deposits were often seen in association with severely affected vessels, such deposits were minimal to absent in HCHWA-D. Even in HCHWA-D patients with one of the most severe degree of CAA, severely stenotic vessels were not shown to be associated with appreciable perivascular plaque pathology. 27

We further observed a different binding pattern of SP subcomponents for histochemical and immunohistochemical reagents (Figure 3, J and K) ▶ . The histochemical dyes such as Masson’s trichrome bound less avidly to the coronal, perivascular, and the primitive plaques, but intensely to the plaque cores, especially to their center-most regions. By contrast, the most peripheral regions of the plaque cores, and also the coronal, perivascular, and primitive plaques, stained strongly with Aβ immunohistochemistry. Such Aβ reactivity pattern is not uncommon for SPs in sporadic or other familial AD.

We studied the association of different plaques in AD/Fl brains with hyperphosphorylated tau (AT8) and reactive glial pathology by double immunohistochemistry. As described earlier, 25,51 CAA in HCHWA-D was associated with only Ubi(+) dystrophic neurites. By contrast, an equal association of both SPs and CAA in AD/Fl brains was observed with not only Ubi(+) dystrophic neurites, but also AT8(+) dystrophic neurites, along with a prominent glial and inflammatory pathology (Table 2 ▶ and Figure 3, D, M, and N ▶ ). Staining AD/Fl brains with angiogenic markers such as vascular endothelial growth factor or basic fibroblast growth factor, a strong reactivity was observed in the vicinity of severely occluded vessels (Figure 3O) ▶ . Diffuse plaques in AD/Fl, similar to those described in other AD as well as in HCHWA-D patients, 2,23 did not consistently associate with a neuritic, glial, inflammatory, or angiogenic pathology (Figure 3P) ▶ .

Morphometric Analysis for AD/Fl Brains

Perivascular plaques, especially those associated with amyloid-laden severely stenotic vessels (ALSSVs) had a striking resemblance to the coronal plaques. We used image analysis to assess sizes of SPs and ALSSVs, the latter defined as when the vessel lumen diameter/CAA diameter was less than or equal to one half. Both ALSSVs and SPs ranged from a few to 600 μm in diameter and the average diameter (±SD) of ALSSVs was 53.4 μm (±43.3) and not significantly different from that of the plaque cores (51.6 ± 48.8 μm, P = 0.3). Furthermore, comparing the ratios of the perivascular and coronal plaque areas to their respective total ALSSV and SP areas, the perivascular plaques constituted 43.2% (±16.4) of ALSSVs, which was not significantly different from the proportion of coronal deposits constituting SPs (49.2 ± 18.3%; P = 0.2) (Figure 4) ▶ .

Predominant Aβ40 Composition of SPs

To identify the precise Aβ species deposited in brains of AD/Fl patients, and also to explore the constitutional similarities of amyloid deposits between AD/Fl, HCHWA-D, and PS1 and sporadic AD patients, serial brain sections were stained with antibodies specific for C-terminal Aβ40 or Aβ42, Aβ N-terminus, and the middle portion of Aβ (Aβ17-24; 4G8). Double immunohistochemistry for these antibodies in various combinations revealed a predominant Aβ(1-40) content of CAA and plaque cores, and a comparable reactivity for Aβ(1-40) and N-truncated Aβ42 in the perivascular and coronal plaques. Diffuse plaques in AD/Fl were entirely composed of N-truncated Aβ42, again resembling diffuse plaques present in the familial and sporadic AD, as well as in HCHWA-D patients. 5,8 Percentage areas of plaques stained by Aβ40 and Aβ42 were studied on serial brain sections using an image analysis program. Aβ40 was the predominant amyloid in vessels in AD/Fl patients (77%), sporadic AD patients (67%), PS1 mutation carriers (60%), and HCHWA-D patients (71%). Aβ40 also constituted a major fraction of Aβ in parenchymal deposits in AD/Fl patients (66%), but only a minor fraction in sporadic and PS1 AD (23% and 26%, respectively) and being completely absent in HCHWA-D patients (Figure 5) ▶ .

Quantification of the relative amounts of Aβ42 and Aβ40 was done by MALDI-TOF mass spectrometry on Aβ immunoprecipitated from frozen AD/Fl brain extracts with 4G8 that does not distinguish between Aβ40 and Aβ42. Full-length Aβ was the major peptide identified in SPs with a 25-fold abundance of Aβ(1-40) greater than Aβ(1-42). Mass spectrometric analysis of amyloid-laden vessels extracted from the brains of Flemish and sporadic AD patients also showed respective 32- and 10-fold higher levels of Aβ(1-40) than Aβ(1-42). However, consistent with earlier observations, 52 analyses of different regions from familial and sporadic AD brains showed that diffuse plaques, but not SPs, were predominantly composed of Aβ(1-42) and Aβ(11-42) (Figure 6) ▶ .

Vasocentric SPs in the Flemish AD

Confocal laser-scanning microscopy using Aβ40- and Aβ42-specific antibodies showed all Aβ40(+) plaques to be positioned around microvessels (positive for CD31, CD34, smooth muscle actin, or C-IV). Three-dimensional reconstructions further displayed a close relationship of amyloid-laden vessels and plaque cores, and the abrupt development of CAA at their points of branching (Figure 7, A and B) ▶ . Light microscopic examination of plastic 1-μm-thick serial sections stained with toluidine blue also demonstrated a close relationship of vessels with plaque cores (Figure 7C) ▶ . Double immunohistochemistry for Aβ and vessel markers revealed the presence of a central or paracentral vessel within 68% of the 2400 SPs studied (Table 3 ▶ ; Figure 8 ▶ ), whereas the remaining were closely associated with the vascular basement membranes of comparatively large-caliber vessels. Most of the primitive plaques enclosed a plaque core or an amyloid-laden vessel in serial section analysis.

Serial ultra-thin section examination revealed a radial arrangement of Aβ fibrils projecting from the basement membrane into the surrounding neuropil as has been classically described for dyshoric angiopathy. 53 With larger vessels, this phenomenon was often limited to only a part of the vessel and core-like compact structures seemed to evolve from the vascular basement membranes (Figure 9) ▶ . The gruel of these larger amyloid deposits or also amyloid deposited abluminally was amorphous, or haphazardly arranged in loose bundles, whereas the amyloid at core-periphery was radially arranged in filamentous bundles. As interpreted from light microscopic studies, the compact amyloid at the center bound more avidly to histochemical stains, whereas the peripheral radial spicules bound more intensely to Aβ antibody; the latter could be because of a relative accessibility of Aβ epitopes.

The noncongophilic diffuse plaques did not have any consistent relationship with vessels, although small amyloid-laden vessels were sometimes noted in these plaque deposits, as has also been observed earlier. 26 Many of these diffuse plaques were also associated with neurons (Figure 3P) ▶ .