Macroscopic pathology
Patient Sw1 (IV:10): ApoE genotype 3/3. The brain weighed 1385 g. Gross examination revealed focal moderate atrophy of parietal superior lobules. There were no signs of infarcts or hemorrhages.
Patient Sw2 (IV:29): ApoE genotype 3/3. The brain weighed 1490 g. A mild degree of atrophy with dilatation of the ventricular system was seen in frontal, parietal and occipital lobes as well as in different parts of the temporal lobe, including gyrus parahippocampalis, hippocampus, and amygdala (Figure 1c). Brainstem and cerebellum had normal macroscopic appearance, apart from mild atrophy of the anterior part of vermis.
Patient Am1 (III:1): ApoE genotype 2/3. The weight of the brain was 822 g. There was a severe degree of atrophy in frontal, temporal and parietal cortices [17].
Patient Am2 (IV:1): ApoE genotype 2/3. After fixation, the brain weighed 1220 g and showed moderate cortical atrophy.
Microscopic pathology of Aβ deposition
The deposition of Aβ in human brain has been suggested to follow a distinct hierarchical sequence, classified as phases 1 to 5 [20]. In the following, we present the structural and immunohistochemical findings in the anatomical regions corresponding to these phases. The pattern of Aβ deposits varied both with respect to the antibodies applied and to the different brain regions analysed. Furthermore, there were some noticeable differences between the four Arctic AD brains studied.
Cerebral neocortices (phase 1)
Histopathology
In H&E stained sections, senile plaques appeared as compact rounded structures with remarkably homogenous texture (Figure 1d). The plaques were devoid of an amyloid core, as shown by the absence of Congo red (Figure 1d, inset) and thioflavin S (not shown) positivity and thus resembled cotton wool plaques (Additional file 1: Figure S9) [2, 3]. With Bielschowsky silver impregnation (without gold enhancement) the plaques were moderately brownish with accentuation of peripheral parts and negative or weakly stained centres, giving the plaques a vaguely ring-like pattern (Figure 1e).
Aβ immunohistochemistry
General features
The area fraction of Aβ immunopositivity (Aβx-42) in frontal, temporal and parietal cortices was about 25%. Plaques were present in all cortical layers, being most numerous and compact in layers 2 and 3, while in deeper layers they were somewhat larger and less distinct. In addition, in layer 1 there were often small diffuse plaques of variable numbers, shapes and staining intensities, as well as thin patchy variably immunopositive subpial bands (Figure 2a-h). The variation in the cortical pattern appeared e.g. in occipital cortex, where layers 4 and 6 were virtually devoid of plaques and layer 5 displayed only a lesser number of diffuse plaques (Figure 2d and Additional file 2: Figure S7c). Variation between patients was also observed e.g. as a paucity of plaques in layer 4 in frontal cortex of patients Am1 and Am2 compared to more abundant plaques in the same region in patients Sw1 and Sw2 (Additional file 3: Figure S1a-g, Additional file 4: Figure S2a, and Figure 2a).
Staining with general Aβ antibodies
With the C-terminal abAβx-42 a majority of neocortical plaques in all four patients were ring-shaped (Figure 2c), as described in the first study on patient Sw1 [17], i.e. the weakly stained centres of larger plaques were surrounded by distinct immunoreactive coronas.
In patient Sw1 the ring pattern visualized with abAβx-42
[11] was also clearly noticeable with abAβx-40, abAβ8–17, abAβ5–10, and abAβarc. The staining was progressively weaker and less distinct with the more N-terminal abs (Additional file 4: Figure S2a-g) and nearly negative with abAβ1–5, although with this antibody the small subpial plaques were still positive (Additional file 4: Figure S2f). No central accentuation with the N-terminal antibodies was observed (Additional file 4: Figure S2e-f).
In patients Sw2 and Am 1, on the contrary, the neocortical plaques displayed a targetoid pattern: In these two brains the plaques were ring-shaped only with the most C-terminal antibody abAβx-42 (Figure 2c), whereas with abAβx-40 the centre was clearly immunopositive (Figure 2b). The mid-domain abAβ17-24 rendered the plaques most compact of all antibodies (Figure 2e), whereas antibodies with epitopes towards the N-terminus stained the plaque coronas more weakly and centres more intensely (Figure 2f-h, Additional file 3: Figure S1d-f). Confocal analysis of sections from patient Sw2 double immunostained with abAβx-42 and abAβ1-5 clearly distinguished the two Aβ components in plaques: the peripheral corona was positive with abAβx-42, while the centre was strongly positive with abAβ1-5 (Figure 3a). Therefore, the plaques could best be described as targetoid, not ring-shaped.
In patient Am2 the staining pattern was more variable. In this patient’s frontal cortex it was similar as in patient Sw1, i.e. the more N-terminal abs also rendered the plaques ring shaped without intensely stained centres, although the staining was less distinct and much weaker (cf. Additional file 4: Figure S2). On the other hand, in Am2 patient’s temporal and occipital cortex many plaques displayed intensely stained centres, similarly to those found in patients Sw2 and Am1 (cf. Figure 2 and Additional file 3: Figure S1).
Staining with specific Aβ antibodies
In patient Sw1 almost all plaques were ring-shaped with the Arctic specific antibody abAβarc (Additional file 4: Figure S2g). In patients Sw2, Am1 and Am2 the staining pattern with abAβarc resembled that with the mid-domain abAβ17-24, although it was somewhat weaker (Figures 3b vs. 2e and Additional file 3: Figure S1g vs. S1c).
In patients Sw2 and Am1 the antibodies abAβ3pE and abAβ11pE showed that N-terminally truncated Aβ starting with pyroglutamate (Aβ3pE and Aβ11pE) colocalized in frontal cortical plaques, although the staining with abAβ11pE was weaker than with abAβ3pE (for Sw2 Figure 3c-d and for Am 1 Additional file 3: Figure S1h-i). Aβ3pE stained most plaques relatively homogeneously and much more extensively than the other N-terminal abAβ1–5 (cf. Figure 2h). Some strongly abAβ1–5 positive centres were also abAβ3pE positive (Additional file 3: Figure S1h). Moreover, we observed vague predominant deposition of Aβ11pE to the plaque centres (Figure 3d).
Correspondence between Aβ immunohistochemistry and mass spectrometry
Analysis by MALDI-TOF of Aβ peptides immunoprecipitated from Sw2 patient’s temporal cortex resulted in various peaks within the m/z range of 2000–5000 Da (Figure 3e and Additional file 5: Table S1). Noteworthy, several of the Aβ species (see labels) corresponded well with the predicted masses of AβpE species, as detected by specific Aβ antibodies (abAβ3pE and abAβ11pE; Figure 3c-d). The observed and predicted m/z values with their relative intensities, corresponding to the peaks shown in Figure 3e are presented in Additional file 5: Table S1.
Allocortical brain regions (phase 2)
Histopathology
Plaques in hippocampus were not as easily discernible with H&E staining as in neocortex–except for those located in dentate gyrus, where they were found to displace granular cells (Additional file 6: Figure S3a). In adjoining occipito-temporal cortex the pattern was similar as elsewhere in cerebral cortex. Although with Bielschowsky silver impregnation, both DNs and NFTs were strongly positive (Additional file 6: Figure S3b and inset), hippocampal plaques were virtually silver negative. However, in the same sections plaques in the nearby occipito-temporal cortex were clearly silver positive and the Aβ immunostainings of hippocampal plaques were intensely positive (see below). Congo red staining was negative (not shown).
Aβ immunohistochemistry
With Aβ immunostaining, the plaques were numerous throughout hippocampus, but their frequency and pattern varied in different hippocampal regions (Figure 4a-f).
In the CA1-CA4 sectors, the general Aβ antibodies disclosed abundant Aβ deposits of variable size and irregular shape. Remarkably, in hippocampus the plaques did not show such a distinct targetoid pattern as in cerebral cortex. Instead, they were small with a diffuse pattern when stained with different Aβ antibodies (insets of Figure 4a-f). In CA3 and CA4 sectors, in addition to the better defined plaques there were also background-like diffuse Aβ deposits, whereas in CA1 and CA2 such deposits were uncommon (Figure 4a-f).
Using the C-terminal abAβx-42 (Figure 4a) and abAβx-40 (Figure 4b), as well as the mid-domain abAβ17–24 (Figure 4c), the staining was recognizably stronger and the plaques were slightly more frequent than with the N-terminal antibodies abAβ8–17, abAβ5–10 and abAβ1–5 (Figure 4b). Diffuse plaques were also abundant in stratum radiatum, subiculum (Figure 4a-f) and transentorhinal cortex, whereas in entorhinal cortex they were relatively sparse. In the adjoining occipito-temporal cortex the plaques appeared similar as elsewhere in neocortex (cf. Figure 2a-h and Additional file 3: Figure S1 and Additional file 4: Figure S2).
Staining of allocortical sections with the specific Aβ antibodies abAβarc (Figure 4e), abAβ3pE (Figure 4f) and abAβ11pE (not shown) gave similar patterns as the general Aβ antibodies. The staining intensities were comparable to that of N-terminal abAβ1−5 (Figure 4d).
Subcortical grey matter nuclei (phase 3)
Histopathology
In basal nuclei, plaques were most often not discernible with H&E or silver staining, but they were selectively positive for Aβ with immunohistochemistry (see below). In this region, the Congo red staining was negative in the parenchyma.
Aβ immunohistochemistry
Among the basal nuclei, claustrum was remarkably deviant: Aβ was deposited as large compact plaques, which had similar targetoid staining pattern as plaques in neocortex (Additional file 7: Figure S4a-g). On the contrary, in the neighbouring putamen the plaques were small and diffusely stained (Additional file 7: Figure S4h-k). These plaques were positive with the C-terminal abAβx-42 (Additional file 7: Figure S4h) and abAβx-40, as well as with mid-domain abAβ17-24 (not shown) and were weakly positive with abAβarc (Additional file 7: Figure S4j) and abAβ11pE (not shown). With the N-terminal abAβ1-5 plaques were almost negative (Additional file 7: Figure S4i), whereas with the pyroglutamate specific N-terminal abAβ3pE the plaques were clearly discernible (Additional file 7: Figure S4k). In amygdala the Aβ deposition was similar to that in putamen, though with C-terminal antibodies the number of small diffuse plaques was greater (Additional file 7: Figure S4l). In thalamus (Additional file 7: Figure S4m) and caudate nucleus (not shown) the plaques were ragged and weakly stained with all antibodies. Globus pallidus was completely negative for Aβ-immunoreactivity (not shown).
Brain stem (midbrain, pons and medulla; phase 4)
Histopathology
In midbrain, pons and medulla, Aβ deposits were not discernible with H&E and only weakly positive with silver staining. None of the parenchymal Aβ deposits were positive for Congo red (not shown).
Aβ immunohistochemistry
In midbrain, the deposition of Aβ was scarce. Diffuse weakly stained Aβ deposits were discernible almost exclusively in nucleus ruber. Among the different Aβ antibodies, only abAβx-42 and abAβ17–24 stained these plaques. However, amyloid angiopathy could be clearly visualized with all Aβ antibodies (not shown). In pons, virtually no parenchymal deposits were observed despite brisk staining of blood vessels (not shown). In medulla, a few distinct plaques, strongly positive with all Aβ antibodies used were present in inferior olivary (Additional file 8: Figure S5a-i) and dorsal vagal nuclei (not shown). The pattern, number and size of plaques in medulla were approximately similar with both general and specific Aβ antibodies, although some variation in the intensity was observed (Additional file 8: Figure S5d-i). Remarkably, abAβx-42 rendered the neuropil in inferior olivary nucleus distinctly positive (Additional file 8: Figure S5a and d), whereas with all other Aβ antibodies it was negative (e.g. Additional file 8: Figure S5b-c and e-i). Olivary neurons within the plaques appeared fairly well preserved (Additional file 8: Figure S5d-k), but both abAβx-42 and abAβ17–24 stained cytoplasmic inclusions within inferior olivary neurons (Additional file 8: Figure S5d-e), as did also PAS and an antibody to lysosomal cathepsin D (Additional file 8: Figure S5j-k; see also paragraph Intracellular Aβ immunoreactivity).
Cerebellum (phase 5)
Histopathology
In H&E stained sections the Aβ deposits were not detectable (not shown). Bielschowsky silver showed no impregnation in the Purkinje cell layer (see below) and only a small number of weakly positive perivascular streaks or smaller deposits in the molecular layer perpendicular to the surface (not shown). As elsewhere, Congo red did not reveal any parenchymal staining.
Aβ immunohistochemistry
The amount of Aβ deposited in cerebellum was much more abundant than that normally seen in AD. Furthermore, the pattern of the deposits was remarkably different from elsewhere in the Arctic AD patients’ brains, especially compared to the cerebral cortices. The immunopositive Aβ deposits were highly variable in size and had very irregular configurations, while distinct rounded Aβ plaques were completely absent. Furthermore, there were marked inter-individual differences, e.g. Aβ deposits in patient Am1 were distinctly different from those in patients Sw1, Sw2 and Am2 (see below).
In patients Sw1, Sw2 and Am2 with similar Aβ staining pattern, the C-terminal abAβx-42 (Figures 5a and 6b) and abAβx-40 (Figure 5b) as well as the mid-domain abAβ17–24 (Figure 5c) displayed diffuse patchy staining in the Purkinje cell layer, from where irregular immunoreactive streaks extended across the molecular layer towards the surface, often following the penetrating blood vessels as wide and irregular cuffs (Figure 5a-c). A similar pattern, although with markedly weaker intensity, was obtained with the pyroglutamate specific abAβ3pE (Figure 5h) and abAβ11pE (Figure 5i). The staining was still weaker with the other N-terminal antibodies abAβ8–17, abAβ5–10 and abAβ1–5 (Figure 5d-f), as well as with abAβarc (Figure 5g). All Aβ antibodies used gave robust staining of blood vessel walls (Figure 5a-i).
In Am1 patient the staining pattern with abAβx-42 and abAβ17–24 (Additional file 9: Figure S6a and c) was almost similar as in the three brains described above, whereas with the C-terminal abAβx-40 (Additional file 9: Figure S6b) and mid-domain abAβ8-17 (Additional file 9: Figure S6d) both the deposits in the granular/Purkinje cell border zone and the streaks in the molecular layer were scarce and weak, although blood vessels were clearly positive. In this brain the more N-terminal abAβ8–17, abAβ5–10 and abAβ1–5 (Additional file 9: Figure S6d-f) stained almost exclusively arterial vessel walls. Among the specific antibodies the staining with Aβarc was faint, whereas both abAβ3pE and abAβ11pE gave clear staining Additional file 9: Figure S6g-i).
Intracellular Aβ immunoreactivity
In all brains definite cytoplasmic immunoreactivity was observed in inferior olivary neurons with abAβx-42 and abAβ17–24 (Additional file 8: Figure S5d-e). The cytoplasmic Aβ(or AβPP [21]) immuno-positivity persisted, even if the formic acid pretreatment was omitted, whereas the extracellular Aβ deposits in the medulla were immunonegative (not shown). These granular cytoplasmic inclusions in inferior olivary neurons were also positive with PAS (Additional file 8: Figure S5j), cathepsin D (Additional file 8: Figure S5k), and α1-antitrypsin (not shown).
Cytoplasmic immunoreactivity with abAβ17–24 was observed also in some other locations, most prominently in cerebellum, both in Purkinje cells (Figure 6a) and in neurons of the dentate nucleus (Figure 6c). Markedly less intense staining was occasionally seen in cerebral cortical and hippocampal pyramidal neurons (not shown).
Microscopy of cellular pathology
General alterations
In all Arctic AD patients’ brains neurons could be frequently identified within the cortical plaques. Interestingly, many of these neurons displayed relatively minor degenerative changes with characteristic vesicular nuclei and prominent nucleoli but had often somewhat condensed cytoplasm (Figure 7a). These neurons were similar as those found within the large cotton wool plaques in PS1Δ9 AD patients (Figure 7b). Moreover, confocal microscopy analysis of cortical sections double-labelled for Aβ and neurofilament showed that some NF-positive axons traverse the plaques (Figure 7c-d). Similarly, Purkinje cells surrounded by abundant Aβ did not appear degenerating, not even those with intracellular abAβ17–24 positive deposits (Figure 6a-b).
Tau immunohistochemistry
Hp-tau immunopositivity was abundant in cerebral cortices and it was mainly present as NTs. The frequency of NTs was accentuated within the Aβ plaques and thus plaques were delineated also by hp-tau staining (Figure 8a-g, Additional file 2: Figure S7a and c). Similar accentuation was also observed in PS1Δ9 AD patients’ cotton wool plaques (Additional file 1: Figure S9i and j). NTs were usually abundant in cortical layers 2 and 3, markedly lesser in layer 4, scarce to absent in layer 5, and slightly accentuated in layer 6. Thus, the frequency of NTs by and large corresponded to that of compact Aβ plaques (Figure 8a-c; Additional file 2: Figure S7a and c). In calcarine cortex, the abundance of NTs corresponded to Braak stage VI, according to the BrainNet Europe recommendation (Additional file 2: Figure S7a). Within Aβ plaques NTs were usually delicate, but occasionally they appeared thicker approaching the appearance of DNs (Figure 8b and g).
Variable numbers of neurons with hp-tau positive NFTs were detected in all cerebral cortical areas examined (Figure 8a, d-e). Neurofibrillary deposits were found in the cytoplasm of small or atrophic neurons (Figure 8a, d-e), whereas larger (non-degenerated) neurons in cortical layers 3, 5 and 6, including those located within Aβ plaques, were usually devoid of NFTs (Figure 8b-c). Most commonly neurons with NFTs were not distributed within though nearby plaques (Figure 8d).
Many hippocampal pyramidal neurons from CA4 to subiculum as well as in adjoining entorhinal, transentorhinal and occipito-temporal cortices harboured prominent NFTs (Figure 8f). Furthermore, many of the granule cells in the dentate gyrus contained distinct hp-tau positive inclusions (Figure 8g). In general, NTs were abundant in the hippocampal neuropil and DNs were more prominent within hippocampal than cortical Aβ plaques (Figure 8g).
In claustrum, the pattern of NTs, DNs and neurons with NFTs was − like that of Aβ plaques − similar as in neocortex (cf. Figure 8a-b and d). In thalamus, where the Aβ plaques were small and diffuse, only few NFTs and very delicate NTs were discernible. In putamen, where plaques were even less conspicuous, almost no NTs were present (not shown). In globus pallidus, where no Aβ deposits were detected, neither were hp-tau immunoreactive structures seen (not shown).
Despite the abundance of Aβ deposits in cerebellar cortex and around Purkinje cells, no neurons (Purkinje cells, granule cells or other neurons) harboured NFTs. In addition, the number of NTs or DNs associated with Aβ deposits was insignificant (not shown).
Macro- and microglial changes
In all four brains, there was slight to moderate reactive astrogliosis in cerebral cortices. GFAP immunopositivity was accentuated within the Aβ-immmunoreactive plaques (Figure 9a-b), mainly as a meshwork of thin processes. Although the number of astrocytic cell bodies in cerebral cortices appeared to be slightly increased, they were relatively diffusely distributed instead of being strictly oriented within or around plaques (Figure 9a-b). Microglial reaction, as determined by Iba1 staining, in cerebral cortices was relatively modest and the distribution of microglial cells did not appear to follow the distribution of plaques (Figure 9b-c). However, their frequency seemed to vaguely correspond to the overall density of the NT meshwork (not shown).
In cerebellum there was pronounced GFAP staining around Purkinje cells (Bergmann astrocytes), although not as prominent as for Aβ. In the molecular layer GFAP positivity did not follow the perivascular streaks of Aβ deposits. Instead, loose GFAP-immunopositive astrocytic network of varying intensity was found throughout the molecular layer (Additional file 10: Figure S8a-b). In all patients, the Iba-1 immunoreactivity in cerebellum was weak and diffuse compared to the Aβ deposition and astroglial reaction and it did not specifically associate with Aβ deposits (Additional file 10: Figure S8c).
Vascular pathology
No major atherosclerosis in the circle of Willis was present, neither were there infarcts, haemorrhages nor microbleeds.
In all four patients’ brains leptomeningeal and cortical penetrating arteries of widely different calibres were variably immunopositive with the different Aβ antibodies used (e.g. Figures 2d and g, 5a-i and Additional file 9: Figure S6a-i). In addition, capillaries in many particular anatomic locations were positive for Aβ, most commonly with abAβx-40 (Figure 2d and Additional file 2: Figure S7c). Presence of fibrillar Aβ within the vessel walls, i.e. evidence of true cerebral amyloid angiopathy (CAA), was verified by Congo positivity with green birefringence in arteries, but not in capillaries (Figure 1d, inset). Details of the composition, distribution and severity as well as illustrations of CAA in these Arctic AD patients’ brains will be presented in a separate article (in preparation).