Modeling familial Danish dementia in mice supports the concept of the amyloid hypothesis of Alzheimer's disease

Abstract

Familial Danish dementia (FDD) is a progressive neurodegenerative disease with cerebral deposition of Dan-amyloid (ADan), neuroinflammation, and neurofibrillary tangles, hallmark characteristics remarkably similar to those in Alzheimer's disease (AD). We have generated transgenic (tg) mouse models of familial Danish dementia that exhibit the age-dependent deposition of ADan throughout the brain with associated amyloid angiopathy, microhemorrhage, neuritic dystrophy, and neuroinflammation. Tg mice are impaired in the Morris water maze and exhibit increased anxiety in the open field. When crossed with TauP301S tg mice, ADan accumulation promotes neurofibrillary lesions, in all aspects similar to the Tau lesions observed in crosses between β-amyloid (Aβ)-depositing tg mice and TauP301S tg mice. Although these observations argue for shared mechanisms of downstream pathophysiology for the sequence-unrelated ADan and Aβ peptides, the lack of codeposition of the two peptides in crosses between ADan- and Aβ-depositing mice points also to distinguishing properties of the peptides. Our results support the concept of the amyloid hypothesis for AD and related dementias, and suggest that different proteins prone to amyloid formation can drive strikingly similar pathogenic pathways in the brain.

Familial Danish dementia (FDD) is an autosomal-dominant disorder caused by a 10 nucleotide duplication just before the stop codon of the BRI2 gene (also known as ITM2B) (1). This mutation generates an extended ORF, resulting in expression of a C-terminally elongated protein. The BRI2-type II integral transmembrane protein is processed by furin-like protease (2, 3), and through this normal processing the FDD mutation results in the release of the longer than normal and amyloidogenic C-terminal cleavage product designated Dan-amyloid (ADan) (represented in Fig. 1A). Similar to many other neurodegenerative diseases, FDD is a disease involving aberrant protein accumulation. Certain features of the disease pathology, such as the accumulation of the ADan peptide as widespread cerebral vascular amyloid and abundant amyloid plaques in the hippocampus, as well as the associated neurofibrillary Tau pathology and neuroinflammation, are remarkably similar to those seen in Alzheimer's disease (AD) (4). The clinical course of disease includes progressive dementia; however, it also features visual cataracts, hearing loss, spasticity, and cerebellar ataxia (4).

Fig. 1.

Age-related deposition of cerebral Dan-amyloid in ADanPP-tg mice. (A) Mutant BRI2 contains a 10-nucleotide duplication just before the stop codon generating a longer ORF (277 instead of 266 amino acids) (1). Normal cleavage at position 243 by a furin-like protease releases the amyloidogenic 34 amino acid long ADan peptide in FDD (2, 3). In addition, BRI2 undergoes processing by ADAM10 to release the extracellular Brichos domain from the membrane-bound N-terminal part that, in turn, undergoes regulated intramembrane proteolysis by SPPL2a/b (42). The recognition sites of various antibodies (Ab) are indicated in green. (B) Western blotting at 2 months of age reveals higher levels of ADanPP in line 7 compared to line 6 (Upper, Ab 1139). Transgene expression in line 7 is several-fold higher than endogenous murine Itm2b (Lower, Ab 1141 recognizes both murine and human BRI2). (C) Western blot of ADanPP7 mouse brain at 2, 4, 10, and 18 months demonstrates that levels of precursor protein remain relatively the same (Upper, Ab 1139) with increasing accumulation of ADan peptide, which runs at the same height as the ADan monomer (M) of FDD patients (Lower, Ab 5282). (D) ADanPP expression in a 2-month-old ADanPP7 mouse (Ab Itm2b). (E) Methoxy staining of ADan lesions in an 18-month-old ADanPP7 mouse. (F–I) Quantification of the amyloid lesions in ADanPP7 mice in hippocampus was done using ADan-specific Ab 5282 in combination with Congo red. At 4 months of age amyloid deposits are seen in the stratum lacunosum moleculare of CA3 (F); at 10 months of age the amyloid appears largely associated with vessels and occurs throughout the dentate gyrus (G); at 18 months of age the entire hippocampus is covered with amyloid lesions (H). Stereological analysis of the Congo-red ADan material (I) revealed a significant increase from 2 to 18 months of age [F(3,24) = 56.404, P < 0.0001] with no gender difference [F(1,24) = 0.017, P = 0.8979; n = 4 females and 4 males, per age group]. (Scale bar, 200 μm.)

The amyloid hypothesis of AD—which may be extended to similar amyloidoses, such as FDD—stresses the role of increased amyloidogenic peptide levels and amyloid formation as a causative factor in instigating disease pathogenesis (5). This hypothesis is strongly supported by the effects of the mutations that lead to familial forms of AD. Indeed, many transgenic (tg) mouse models of AD recapitulating cerebral β-amyloidosis (Aβ) and associated lesions, neuroinflammation, and memory impairment, are based on familial AD mutations that increase Aβ production (6–9). However, the majority of AD cases are sporadic and challenges have been raised to the amyloid hypothesis by arguing that the increase in Aβ plaque pathology and neurofibrillary tangle formation may be linked through parallel but independent mechanisms (10). Additionally, Aβ deposition and tangle formation may be disease bystanders rather than causative for the disease (11).

To advance our understanding of FDD pathogenesis, we generated tg mice that overexpress the Danish mutant form of BRI2(ADan precursor protein; ADanPP). By cross-breeding these mice, we have examined the relationship between ADan- and Aβ pathology and the impact of ADan on the promotion of Tau pathology.

Results

ADanPP-Tg Mice Demonstrate the Age-Related Deposition of ADan.

Tg mice expressing the Danish mutant form of BRI2 (Fig. 1A) were generated on a C57BL/6 background using the cosmid-based Syrian Hamster prion protein expression cassette. Two tg lines were selected (ADanPP6 and ADanPP7). The ADanPP7 line revealed higher levels of both full-length and cleaved precursor protein expression compared with the ADanPP6 line, and several-fold that over endogenous murine Bri2 (Fig. 1B). The ADanPP7 mice were then analyzed more extensively.

Precursor protein expression in ADanPP7 mice remains stable with aging; however, ADan peptide is initially detected by routine Western blotting at 4 months of age and increases dramatically with aging (Fig. 1C). Immunohistochemical analysis revealed ADanPP expression throughout the brain, mainly in neurons and to a lesser degree in astrocytes, with the strongest staining in hippocampus and neocortex (Fig. 1D). As early as 2 months of age, ADan deposition, detected by immunohistochemistry or amyloid-binding dyes (thioflavin S, Congo red, methoxy), was observed in hippocampus and in meningeal vessels. At 18 months of age, ADan deposition occurs throughout the brain (Fig. 1E), including the amygdala, thalamus, brainstem, and to a lesser extent the cerebellum, with the majority of ADan-deposits associated with the vasculature. Stereological quantification in the hippocampus of 2-, 4-, 10-, and 18-month-old ADanPP7 mice confirmed the age-related increase in ADan deposition, which starts in the stratum lacunosum moleculare of the CA3 region, followed by the dentate gyrus and the CA1 region. ADan accumulation did not significantly differ between genders (Fig. 1 F–I). ADan deposition in the ADanPP6 line showed a delayed progression compared to the ADanPP7 line, but generally followed a similar distribution.

ADan Lesions and Their Impact on the Neuropil and Vasculature.

ADan deposition was primarily associated with the vasculature. Although in larger vessels the ADan was confined to the vessel wall with a sheet-like appearance (Fig. 2A), other smaller vessels revealed a thick coat of ADan that often completely obstructed the vessel lumen (Fig. 2B). Perivascular plaques often surrounded a significant portion of the vessel surface, and appeared as parenchymal plaque-like structures (Fig. 2C). Some lesions appearing to be parenchymal ADan deposits (Fig. 2D) were typically surrounded by a cloud of small punctate ADan-immunoreactive, but Congo red-negative, aggregates. By 18 months of age, entire regions—including the hippocampus, cortex, and brainstem—were covered with diffuse clouds of the small punctate ADan-immunoreactive aggregates.

Fig. 2.

ADan lesions and their impact on the neuropil and vasculature. (A–D) Various types of ADan lesions in 18-month-old ADanPP7 tg mice (Ab 5282) at light microscopic level. (E–G) Ultrastructural appearance of the ADan lesions. Amyloid fibrils (asteriks), microglia cell nucleus (N); dystrophic neurites (DN); microglia cell with the typical features of phagosomal/lysosomal material (arrowhead in G). (H and I) Confocal microscopy of double-immunolabeled vessels (red, smooth muscle-cell actin; green, ADan) in a non-tg control (H) and tg (I) mouse. Note the focal disappearance of smooth muscle cells at sites of ADan deposition (arrowheads, I). (J) Hemosiderin-positive microglia (blue) reveal the occurrence of cerebral microhemorrhages. (K) Hemorrhage frequency per hemisphere in tg and non-tg 18-month-old mice [F(1,14) = 14.440, P < 0.002; n = 8, four females and four males per group). (L) Dystrophic synaptophysin-positive structures in the vicinity of both parenchymal and vascular ADan deposits (arrowhead), but also throughout the parenchyma (arrow). (M and N) Cresyl violet staining of the hippocampus of an 18-month-old tg mouse (N) compared with a non-tg littermate (M). (O) Number of CA3 neurons in 18-month-old ADanPP7 mice did not reveal significant changes [F(1,10) = 0.326, P = 0.5806; n = 6 non-tg, 6 tg; five males, seven females). [Scale bars: 20 μm (A–D, J, L), 2 μm (E, G), 3 μm (F), 10 μm (H, I), 200 μm (M, N).]

Ultrastructural analysis revealed that ADan is highly compact in nature (Fig. 2 E–G). Amyloid in the vessel wall appeared to be integrated into the endothelial and vascular basement membrane and lead to an abnormal thickening of the membrane and destruction of vessel wall integrity (Fig. 2E). Often, such perivascular amyloid penetrated into the surrounding parenchyma and was found to be surrounded by microglia and dystrophic neurites (Fig. 2F). Smaller “plaque-like” lesions without an obvious vascular component were also observed, typically surrounded by microglial cytosolic structures (Fig. 2G).

With age, the accumulation of ADan in the vasculature led to a loss of vascular smooth-muscle cells (Fig. 2 H and I). Consistent with a weakening of the vessel wall, vessel-associated microbleeds were observed in the 18-month-old ADanPP7 tg mice, although no incidence of microhemorrhage was found in the control non-tg mice (Fig. 2 J and K).

Associated primarily with parenchymal ADan lesions, dystrophic boutons were detected (Fig. 2L), an observation consistent with the ultrastructural analysis (Fig. 2F). Additionally, structures consistent with (secondary) axonal swelling or spheroids were observed throughout the brain in regions with ADan deposition (Fig. 2L). No significant neuron loss was found in the hippocampus, although there was an apparent displacement of neuronal cell bodies in the CA3/4 region of the hippocampus (Fig. 2 M–O).

ADan Lesions Evoke a Neuroinflammatory Response.

ADan deposition elicited a robust activation of astrocytes, revealed by darkly stained GFAP-immunoreactive hypertrophic cell bodies and processes (Fig. 3 A–F). Astrocytosis, initially seen in the stratum lacunosum moleculare of the CA3 hippocampal region (Fig. 3D), spread throughout the hippocampus and neocortex, and closely followed the temporal pattern of ADan deposition (Fig. 3E). By 18 months of age, astrogliosis was seen throughout the entire brain (Fig. 3F). Similarly, microgliosis closely followed the spreading and temporal development of the amyloid lesions (Fig. 3 G–L). Although activated astrocytes covered entire areas containing ADan deposits with hypertrophic processes (Fig. 3M), activated microglia were observed in close juxtaposition to amyloid deposits, with one to four microglial cells per ADan deposit (Fig. 3N). Confocal microscopy confirmed the tight association between microglia and ADan deposits (Fig. 3O).

Fig. 3.

Glial activation in response to ADan deposition. (A–F) GFAP immunohistochemistry reveals an activation of astrocytes (arrowheads) at sites of Congo-red stained ADan deposition (red) in 4-month-old (D), 10-month-old (E), and 18-month-old (F) ADanPP7 tg mice. No such hypertrophic and darkly stained astrocytes are seen in non-tg littermates (A, B, and C, respectively). (G–L) Similarly, activation of Iba1-positive microglia in ADanPP7 mice closely follows the temporal-spatial pattern of ADan deposition (arrowheads in J, K, and L). Corresponding non-tg littermates in (G, H, and I, respectively). (M and N) Higher magnification of GFAP⁺ astroglia (M) and Iba1⁺ microglia (N) with congophillic amyloid (red). (O) Confocal imaging of Iba1 (red) and amyloid (green; methoxy staining) gives the impression that microglia are capable of engulfing ADan (shown is a z-stack of 28 slices). [Scale bars: 300 μm (A–L), 50 μm (M and N), 5 μm (O).]

ADanPP Mice Have Behavioral Deficits, Including Increased Anxiety.

In the place and cue navigation task of the Morris water maze, 18- to 20-month-old ADanPP7 mice took longer to find the platform compared with age-matched non-tg controls (Fig. S1A). When the swim path was analyzed, a flattened learning curve was found (Fig. S1B). Thigmotaxis and passive floating were increased in the tg mice; however, swim speed was reduced (Fig. S1). During the probe trials, tg mice tended to have a lower preference for the trained target zone compared with the controls (Fig. S1 C and D). Searching precision during the two probe trials revealed a minor reduction after 24 h, and a total loss of spatial selectivity after 6 d (Fig. S1 E and F).

In the open field, the ADanPP7 mice revealed reduced activity in the beginning of each of the two observation periods (Fig. S1G). This observation, together with the increased preference of tg mice for the wall zone (Fig. S1H), the observed reduction of vertical activity, and the more numerous fecal boli in the open field (Fig. S1), is indicative of an anxiety-related phenotype of the ADanPP7 mice.

At 6 months of age, ADanPP7 mice did not display the phenotypic changes that were observed in the aged mice, suggesting that the behavioral changes in the ADanPP7 mice are age-dependent (Fig. S1).

Analysis of body weight revealed that ADanPP7 tg mice failed to gain weight with age (Fig. S1I). This failure to continue to gain weight is temporally coincident with brain amyloid accumulation. By 18 months of age, the ADanPP7 mice were 20 to 30% lighter than their non-tg littermates, and additionally showed alopecia and kyphosis.

Aβ Deposition Is Reduced and Not Coincident with ADan Deposition in Double-Tg ADanPP7/APPPS1 Mice.

The Aβ and ADan peptides have been reported to colocalize in amyloid lesions of FDD patients, suggesting a disease-promoting pathogenic interaction (4, 12, 13). To examine this, ADanPP7 mice were crossed with Aβ-depositing APPPS1 tg mice (14) and analyzed at 4 months of age, when single-tg mice exhibit ADan and Aβ deposits, respectively. Surprisingly, double-tg ADanPP7/APPPS1 mice showed a 68% reduction in neocortical Aβ deposition compared with APPPS1 single-tg littermates (Fig. 4 A–C). Western blot of whole-brain homogenates confirmed a decrease in both Aβ40 and Aβ42 in double-tg mice compared with single-tg APPPS1 mice, while levels of amyloid precursor protein (APP) remained relatively unchanged (Fig. 4 D–F). In contrast to Aβ, ADan levels were not different between double-tg ADanPP7/APPPS1 and single-tg ADanPP7 littermates (Fig. 4 G–I). Immunohistochemical analysis of ADan deposition also showed no appreciable difference between double-tg ADanPP7/APPPS1 and single-tg ADanPP7 mice. Given that BRI2 expression has been shown to alter APP metabolism and Aβ generation (15–18), brain Aβ levels were determined in 1.5-month-old predepositing mice. A small, but not statistically significant, decrease (15–17%) in both Aβ40 and Aβ42 was seen comparing double-tg mice with single-tg APPPS1 littermates [Aβ40: 0.72 ± 0.21 and 0.60 ± 0.08 pmol/g wet weight; F(1,14) = 0.466, P = 0.506; Aβ42: 0.97 ± 0.19 and 0.82 ± 0.09 pmol/g; F(1,14) = 0.686, P = 0.421; n = 5 single tg, 11 double tg; all females]. To confirm the reduction in Aβ pathology by ADanPP/ADan and to eliminate any confound because of the high Aβ42-drive from the PS1 mutation in the APPPS1 line, the lower-expressing ADanPP6 line was crossed with the APP23 mouse model (9). At 13 months of age, again when single-tg mice exhibit ADan and Aβ deposits, a 70.0% reduction in Aβ deposition in the neocortex of double-tg ADanPP6/APP23 mice was seen compared to single-tg APP23 mice [0.42 ± 0.11% vs. 1.41 ± 0.42% Aβ load; n = 6 and 4 per group; all females; F(1,8) = 7.734, P < 0.03].

Fig. 4.

Aβ deposition is reduced and not coincident with ADan deposition in double-tg ADanPP7/APPPS1 mice. (A and B) Aβ immunostaining of 4-month-old ADanPP7⁺/APPPS1⁺ double-tg mice compared with single-tg ADanPP7⁻/APPPS1⁺ littermates reveals a decrease in Aβ deposition. Higher magnifications additionally show a change in plaque morphology (i.e., less but occasionally larger plaques in the double-tg mice). (C) Quantification of neocortical Aβ load reveals a drastic decrease in Aβ deposition in the double-tg mice [F(1,10) = 17.852, P < 0.0019; all females; n = 5 single tg, n = 7 double tg]. (D and E) Western blotting of APP and Aβ40/42 (Ab 6E10) in ADanPP7⁺/APPPS1⁺ mice in comparison with ADanPP7⁻/APPPS1⁺ siblings reveals no change in APP but a decrease in the double-tg mice in both Aβ40 and Aβ42. Shown are three mice from each genotype. (F) Densitometry of Aβ levels of all of the mice confirms the significant decrease in both Aβ40 [71.4%, F(1,10) = 7.444, P < 0.0213] and Aβ42 [47.9%, F(1,10) = 8.106, P < 0.0173] species (all females; n = 5 single tg, 7 double tg). (G and H) Western blotting of ADanPP (Ab Itm2b) and ADan peptide (Ab 5282) reveals no change in ADanPP and ADan levels between genotypes. Shown are three mice each. (I) Densitometry of ADan levels of all of the mice indicates no significant difference (all females; n = 4 single tg, 7 double tg). (J–M) Confocal microscopy reveals deposition of both peptides (green, ADan; red, Aβ) in some amyloid plaques in the neocortex of double-tg mice; however, plaques consistently contained separate amyloid foci in which the two proteins do not colocalize (J). Other amyloid deposits in the hippocampus demonstrated deposition of only ADan or only Aβ in close vicinity to each other (K). Plaques consisting of only Aβ were often decorated with ADanPP/ADan⁺ dystrophic boutons (arrowheads in L). Vascular amyloid consistently contained either ADan alone, or fascinatingly, both ADan and Aβ as separate foci along the same vessel, without colocalization. [Scale bars: 500 and100 μm (A and B), and 20 μm (J–M).]

Double immunolabeling of sections from both ADanPP7/APPPS1 mice and ADanPP6/APP23 mice showed limited coincidence of the two amyloids. Although ADan and Aβ appeared to coexist in individual amyloid plaques, they formed separate foci within a given plaque (Fig. 4J). More interestingly, a large portion of plaques consisted of only one amyloid peptide (Fig. 4 K and L). In vessels showing both ADan and Aβ deposition, ADan and Aβ were localized to separate areas along the same vessel, with little or no colocalization (Fig. 4M).

ADan Accumulation Accelerates Tau Pathology.

The occurrence of neurofibrillary Tau pathology in combination with ADan deposition in FDD (4) suggests that ADan and Tau may participate in a common neurodegenerative pathway. To test this experimentally, ADanPP7 mice were crossed to TauP301S tg mice (19). Mice were aged for 12 months, an age at which single TauP301S tg mice have robust Tau lesions in the brainstem and only infrequent lesions in the neocortex (19). Double-tg ADanPP7/TauP301S mice had a >50-fold increase in Gallyas-positive neurofibrillary pathology in the neocortex compared with the single-tg TauP301S littermates (Fig. 5 A and B). When phosphorylated AT8⁺ Tau inclusions were counted, an 18-fold increase in neocortex was found in double-tg mice compared with single-tg TauP301S littermates [627.0 ± 277.6 vs. 33.4 ± 13.8; F(1,11) = 5.396, P < 0.04; n = 6-7/group; all females]. Western blotting of sarcosyl-insoluble Tau confirmed the increase in higher molecular weight insoluble Tau species (64 kDa band) in the forebrain of double-tg mice compared with Tau single-tg siblings (Fig. 5C). Immunohistochemistry and Western blotting for ADan did not reveal any changes between double-tg mice and single-tg ADanPP7 littermates (Fig. 5 D–F). Finally, predepositing 1.8- to 2-month-old double-tg ADanPP7/TauP301S mice did not reveal any AT8⁺ Tau inclusions and no difference was noted in AT8-immunoreactivity between the double-tg and single-tg TauP301S mice.

Fig. 5.

ADan accelerates the occurrence of Tau lesions in double-tg ADanPP7/TauP301S mice. (A) Gallyas silver staining of 12-month-old ADanPP7⁺/TauP301S⁺ mice reveals increased Tau lesions in both neocortex and brainstem in comparison with single-tg ADanPP7⁻/TauP301S⁺ littermates, with no staining in ADanPP7⁺/TauP301S⁻ mice. Inserts are higher magnifications. (B) Quantification of neocortical Gallyas-positive cells shows a dramatic increase in ADanPP7⁺/TauP301S⁺ mice [all female; n = 6 double tg, n = 7 single tg; F(1,11) = 5.377, P < 0.041]. (C) Western blotting of the sarcosyl insoluble fraction of forebrain “F” and hindbrain “H” homogenates using Ab HT7 reveals an increase in the forebrain of ADanPP7⁺/TauP301S⁺ mice (Lane 3) in comparison with ADanPP7⁻/TauP301S⁺ littermates (Lane 1), while the increase in the hindbrain appears smaller (Lanes 4 and 2, respectively). (D and E) No difference in neocortical ADan pathology (Ab 5282) between ADanPP7⁺/TauP301S⁺ mice and ADanPP7⁺/TauP301S⁻ littermates is observed (D) consistent with the stereological quantification [E; F(1,10) = 1.014, P = 0.338; all females; n = 6 double tg, 6 single tg]. (F) Western blotting also does not reveal a difference in ADan peptide levels between ADanPP7⁺/TauP301S⁺ mice (Lanes 3 and 4) in comparison with ADanPP7⁺/TauP301S⁻ littermates (Lanes 5 and 6). (G) Phenotyping of the Tau-lesions in the neocortex of ADanPP7⁺/TauP301S⁺ mice reveals a Tau conformation folded at the N terminus (Ab Alz-50) with mostly an intact C terminus (Ab Tau-46.1). The Tau lesions are Thiazin red (TR)-negative, while ADan deposits in the wall of blood vessels are TR-positive (see merger of Alz-50, Tau46.1, and TR staining). The lesions consist of hyperphosphorylated Tau at Ser396-404 (Ab PHF1) and Ser199-202 (Ab pSer199-202). For comparison, phenotyping of the Tau lesions in the neocortex of 12-month-old APPPS1⁺/TauP301S⁺ mice reveals the same conformation and phosphorylation pattern. For a list of antibodies and stains, see Table S1. [Scale bars: 100 and 20 μm (A–F), 10 μm (G).]

The potentiation of Tau lesions in aged double-tg ADanPP7/TauP301S mice compared with single-tg Tau mice is in all aspects reminiscent of previous studies reporting increased Tau pathology in Tau tg mice crossed with lines that develop Aβ plaque pathology (20–22). To directly compare the ADan-induced Tau lesions to Aβ-induced Tau lesions, APPPS1 mice were crossed with the TauP301S mice and analyzed at 12 months of age. Again, an increase (3- to 4-fold) in AT8⁺ Tau lesions was observed in the neocortex of double-tg APPPS1/TauP301S mice compared with single-tg TauP301S mice (212 ± 53 vs. 62 ± 50, respectively; n = 8–9/group; males and females combined; P = 0.05). Subsequent comparison of Tau lesions between ADanPP7/TauP301S mice and APPPS1/TauP301S mice by a panel of antibodies against conformationally altered, phosphorylated, and truncated forms of Tau (Table S1) revealed no differences, consistent with the similar morphological and regional appearance of the Tau lesions in the two crosses (Fig. 5G).

Discussion

The identification of mutations in BRI2 causing familial British dementia (FBD) and FDD (1, 23) has stimulated intense research into the pathophysiology of these rare diseases. This research interest is sparked because of the similarities between the neuropathological lesions in FBD and FDD and those of AD (4, 12). Modeling FBD and FDD in mice was initially disappointing, but recent work has begun to successfully model FDD in mice (24, 25).

The ADanPP mouse model described here recapitulates the cerebral amyloidosis of FDD. In both ADanPP mice and FDD patients, ADan lesions are abundant in the hippocampus, and also occur in the neocortex (4). The cerebellum in FDD appears to be more affected compared with the mouse model; however, this may be in part a result of the promoter used to drive trangene expression. Most of the ADan lesions in the ADanPP mouse model are associated with the vasculature, as is the case for FDD patients (4). This finding is striking, given that a primarily neuronal promoter was used for the expression of ADanPP. Tg mouse models of cerebral β-amyloidosis using the same PrP promoter accumulate Aβ primarily in the brain parenchyma in form of amyloid plaques (6, 8). These findings argue that properties of the amyloidogenic peptide, in this case Aβ versus ADan, and not the promoter used, can determine in which compartment amyloid formation occurs in the brain (26).

Neuroinflammation is prominent in FDD and in ADanPP mice, with microglia surrounding congophilic ADan deposits and extensive astrocytosis (4). Similarly, dystrophic boutons associated with parenchymal and perivascular amyloid plaques, and amyloid-associated vessel wall degeneration with loss of smooth muscle cells and hemorrhage are seen in both FDD and the ADanPP mouse model (4). All these amyloid-associated changes are remarkably similar to the neurodegeneration, neuroinflammation, and vascular pathology described in AD and in mouse models of cerebral β-amyloidosis (6–9, 14, 26–29). Thus, the shared pathophysiology of ADan and Aβ accumulation suggests that, regardless of protein sequence, the amyloidogenic properties of the ADan and Aβ peptides are responsible for driving similar downstream disease pathologies.

A common mechanism of amyloid pathophysiology, independent of the sequence of the amyloidogenic peptide, is further supported by our findings that ADan, like Aβ, can induce Tau pathology. Tau lesions in ADanPP mice are absent unless ADanPP mice are crossed with a mouse model overexpressing human mutant Tau. The same is true for APP and APP/PS tg mouse models of AD (20–22). When crossed, the promotion of tau phosphorylation and aggregation appears remarkably similar in ADanPP/TauP301S and in APPPS1/TauP301S mice, and mimics in many aspects human neurofibrillary tangle formation in FDD and AD (4, 30, 31). Thus, our findings suggest that Aβ and ADan (or their respective precursors) share a similar mechanistic link leading to the induction of Tau pathology, possibly mediated by GSK-3 or ChIP (22, 32). The importance of endogenous Tau for Aβ-mediated dysfunction has been recently demonstrated (33), and this may contribute to the behavioral dysfunction of ADanPP mice as well.

However, Aβ and ADan also have distinguishing properties within the brain. The lack of increased amyloid deposition and codeposition in the ADanPP/APP tg mice was unexpected in light of the reported colocalization—and suggested disease-promoting effect—of Aβ and ADan in FDD. However, in patients the colocalization of Aβ and ADan has been reported to be incomplete (4, 12, 13), as was seen in the mice. Our observations of reduced Aβ deposition in the ADanPP/APP tg mice are consistent with previous studies, showing that secreted and potentially amyloidogenic peptides, such as Aβ1–40, Bri2-23, cystatin C, and transthyretin, can inhibit Aβ pathogenesis (34–37). These peptides have been reported to bind Aβ, suggesting that the formation of heterogenous structures may prevent mature amyloid fibril formation (35, 38, 39). Indeed, for cystatin C, the inhibition has been shown to be bidirectional, with Aβ also inhibiting cystatin C dimerization in the brain (35). Understanding the interdependence of Aβ and ADan deposition, however, is complicated by the potential role BRI2 can play in inhibiting APP processing through binding of its transmembrane and Brichos domain to the Aβ region of APP (15–18). Although only a nonsignificant 15 to 17% decrease in Aβ was found in predepositing double-tg ADanPP/APP mice compared with single-tg APP mice, such a steady-state reduction in Aβ generation may be sufficient to account for the observed significant reduction of Aβ deposition several months later. Indeed, that ADan deposition was not affected in the ADanPP/APP crosses is consistent with recent results suggesting that only C-terminal cleaved BRI2 affects APP processing and Aβ generation (18, 40).

A difference between ADanPP mice and mouse models of cerebral β-amyloidosis is the reduced body weight, and (presumed) early death of ADanPP mice (tg mice beyond 24 months of age could not be included in our analyses because mice had to be killed as a result of the significant rate of morbidity). Premature death has only been reported in a few APP and APP/PS1 tg lines and does not appear to be related to the amyloid pathology (6, 33, 41). Although the temporal convergence of ADan deposition and the body-weight loss make a causal link possible, alopecia and kyphosis may indicate that ADanPP expression outside the CNS is responsible for these phenotypes.

Our findings argue that Aβ and ADan, in spite of their lack of sequence homology, participate in similar pathways, leading to neuroinflammation and dementia through their shared amyloid properties. Because ADan does not occur in healthy individuals, our results imply that the relationship between Aβ and AD pathogenesis is likely not directly related to Aβ physiology. Extending our understanding of such shared pathways will lead to the identification of mechanistic targets for therapeutic intervention common to apparently disparate, albeit pathophysiologically related, neurodegenerative diseases.

Materials and Methods

Generation of ADanPP-Tg Mice.

cDNA constructs encoding human BRI2 with the Danish mutation were microinjected into C57BL/6N pronuclei (C57BL/6N-Tg(SHaPrP-BRI2_{795InsTTTAATTTGT}). Founders were bred with C57BL/6J mice. For cross-breeding, APPPS1-21 mice (14), APP23 mice (9), and TauP301S mice (19) were used. For details, see SI Materials and Methods.

Tissue Preparation, Histology, and Stereology.

Formaldehyde-fixed free-floating or paraffin-embedded serial sections were used. For details and antibodies, see SI Materials and Methods.

Western Blot and Aβ Immunoassay.

For Western blotting, various previously described gel systems were used. Human Aβ40/42 was measured by sandwich immunoassay using MULTI-ARRAY Human (6E10) Aβ ultra-sensitive kits (Meso Scale Discovery). For details see SI Materials and Methods.

Behavior Experiments.

Morris water-maze testing was done in a circular pool (150 cm diameter). Open-field testing was done in a round open-field (150 cm diameter). All trials were tracked using a Noldus EthoVision system (Noldus Information Technology). Raw data were transferred to the public domain software Wintrack2.4 to calculate the various parameters. For details see SI Materials and Methods.