Exogenous Induction of Cerebral ß-Amyloidogenesis Is Governed by Agent and Host

See allHide authors and affiliations

Science  22 Sep 2006:
Vol. 313, Issue 5794, pp. 1781-1784
DOI: 10.1126/science.1131864


Protein aggregation is an established pathogenic mechanism in Alzheimer's disease, but little is known about the initiation of this process in vivo. Intracerebral injection of dilute, amyloid-β (Aβ)–containing brain extracts from humans with Alzheimer's disease or β-amyloid precursor protein (APP) transgenic mice induced cerebral β-amyloidosis and associated pathology in APP transgenic mice in a time- and concentration-dependent manner. The seeding activity of brain extracts was reduced or abolished by Aβ immunodepletion, protein denaturation, or by Aβ immunization of the host. The phenotype of the exogenously induced amyloidosis depended on both the host and the source of the agent, suggesting the existence of polymorphic Aβ strains with varying biological activities reminiscent of prion strains.

The accumulation of misfolded proteins is a common feature of several neuro-degenerative disorders. In Alzheimer's disease (AD), the multimerization of the Aβ peptide is an early and central process in the pathogenic cascade (13), but little is known about the mechanisms that govern the initiation of Aβ aggregation and deposition in vivo. Ordered protein aggregation in vitro is a function of protein concentration and time and follows a crystallization-like polymerization mechanism that can be rapidly initiated by introducing an exogenous seed (4). In vivo, seeded aggregation of Aβ is seen after injecting AD brain extracts into the brains of nonhuman primates (5) or APP-transgenic mice (6), reminiscent of the conformational conversion mechanism of prion infectivity (79).

We injected 10% (w/v) extracts of brain homogenates from autopsied AD patients (AD extract) or from aged, β-amyloid–laden APP23 transgenic mice (10) (APP23 Tg extract) into the hippocampus of young male APP23 mice. Four months later, the host mice were analyzed (11). Both AD extract and APP23 Tg extract induced robust deposition of Aβ in the hippocampus (Fig. 1, A and B). Intracerebral injection of tissue extract from an aged control patient induced only minimal Aβ deposits, consistent with the low Aβ load in the donor (Fig. 1C; fig S1). No seeded Aβ deposits were found after control injections of brain extract from an aged, wild-type mouse or of phosphate-buffered saline (PBS) (Fig. 1, D and E). Infusion of APP23 Tg extract into wild-type mice did not induce Aβ deposition; thus, the observed Aβ deposits did not simply represent the injected Aβ-containing material (Fig. 1F), and imply that host factors are critical for in vivo seeding. No seeded Aβ deposits were observed when APP23 Tg extract from a young, 2-month-old, predepositing mouse was injected into APP23 hosts. The concentrations of Aβ in the AD and APP23 Tg extracts were estimated to be 1 to 10 ng/μl. In both AD extract and APP23 Tg extract, Aβ monomers, oligomers, and larger multimeric species were present (Fig. 1H).

Fig. 1.

Brain extract (10%) was injected into the hippocampus of 5-month-old male APP23 hosts (A to E) and nontransgenic littermates (F). Mice were analyzed 4 months later. Injection of AD extract (A) and APP23 Tg brain extract (B) induced numerous Aβ-immunoreactive deposits. Few or no Aβ deposits were detected after injections of brain extract from an aged (95 years) control (Ctrl) patient (C) or wild-type (Wt) mouse (D). No Aβ deposits were observed after PBS injections (E) or when Tg extract was injected into wild-type mice (F). Stereological quantification of Aβ load by immunohistochemistry (G) confirmed significant amyloid induction by AD and Tg brain extracts compared to Ctrl and Wt extracts (n = 5 mice per group; mean ± SEM, P < 0.001). (H) Tris-Tricine SDS–polyacrylamide gel electrophoresis (PAGE) followed by immunoblotting with a human Aβ-specific antibody. Lane 1: synthetic Aβ1-40 + Aβ1-42 (1 ng/μl each); lane 2: AD brain extract; lane 3: Ctrl brain extract; lane 4: APP23 Tg brain extract; and lane 5: Wt brain extract. For each extract, 1 μl was loaded. Arrowheads indicate monomeric, dimeric, trimeric, and tetrameric Aβ. Aβ concentration in the control patient extract was below the detection level. When less dilute samples were used, Aβ was detected in the Ctrl extract, consistent with the sparse amyloid plaques in this patient (fig. S1) and the modest seeding activity of the Ctrl extract. Scale bar, 350 μm.

The localization and biochemical nature of the induced Aβ lesions were markedly similar to those seen in normal, aged APP23 transgenic mice (10, 12). Induced Aβ deposits were found primarily in the injected hippocampus, but some were also observed in the dorsal lateral geniculate nucleus, corpus callosum, and entorhinal cortex, and in the vasculature of the thalamus and pia mater (fig. S2). Most induced Aβ deposits were diffuse, although some congophilic amyloid plaques were present and surrounded by activated microglia, astrocytes, and dystrophic neurites (fig. S2). Aβ in micropunches taken from the hippocampus of injected APP23 mice was mainly parechymal and consisted of Aβ1-40 and Aβ1-42, whereas the amyloid in the thalamic micropunches was mostly vascular and consisted predominantly of Aβ1-40 (fig. S3).

The exogenous induction of Aβ deposition in vivo also is time and concentration dependent (fig. S4). Immunoreactive Aβ deposits first appeared 2 months after injection and thereafter increased significantly with time. APP23 extract that was further diluted to 0.5% produced a pattern of deposition in the host similar to that seen with the 10% extract, but with much less potency (fig. S4).

The phenotype of exogenously induced β-amyloid deposits is dependent on both the agent and the host. We injected brain extract from aged APP23 mice into APP–presenilin-1 (PS1) transgenic hosts, and vice versa (Fig. 2). Amyloid deposition in the hippocampus of male APP23 mice normally begins at 9 to 10 months of age, whereas APPPS1 mice develop Aβ deposition in the hippocampus at 3 to 4 months of age (10, 13). Total Aβ concentration was similar in the two extracts (Fig. 2I), but in APPPS1 extracts, the highly amyloidogenic Aβ1-42 was several times as abundant as Aβ1-40, whereas in extracts from APP23 mice, more Aβ1-40 was present. The amyloid-inducing activity of APP23 and APPPS1 extracts was quantitatively similar 3 months after infusion, but injection of APPPS1 extract into APP23 hosts consistently induced a coarse pattern of compact, punctate Aβ deposition that was mainly confined to the subgranular layer of the hippocampus, whereas the APP23 extract injected into APP23 hosts yielded primarily diffuse and filamentous lesions, with substantial diffuse Aβ in the molecular layer (Fig. 2, A to D). When the same extracts were injected into the APPPS1 host, lesions induced by the APPPS1 extract were even more coarse and punctate, whereas those induced by the APP23 extract were a mixture of filamentous and compact types (Fig. 2, E to H). In mice analyzed 1 month after the infusion of the extract, appreciable coarse Aβ induction was already apparent in APPPS1 hosts but not in APP23 hosts.

Fig. 2.

Brain extract (10%) from an aged APPPS1 mouse or an aged APP23 mouse was injected intrahippocampally into either 5-month-old male APP23 or 2-month-old male APPPS1 hosts. Brains were immunohistochemically analyzed for Aβ 3 months later. The panels to the left (A, C, E, and G) show overviews of the hippocampus, and the panels to the right (B, D, F, and H) show corresponding higher magnification images of the upper blade of the dentate gyrus (ML, molecular layer; GL, granular cell layer; SGZ, subgranular cell layer). The coarse, punctate pattern of Aβ staining in the APP23 host is apparent, particularly in the SGZ, after injection of the APPPS1 extract (A and B). Injection of APP23 extract induced more filamentous and diffuse Aβ in the ML (C and D). When APPPS1 extract was injected into the APPPS1 host, the coarse, punctate pattern of Aβ induction was even more distinct (E and F), whereas the APP23 extract in APPPS1 hosts induced Aβ lesions intermediate in appearance to the coarse and filamentous types (G and H). At 5 months of age, APPPS1 hosts had developed some Aβ deposits endogenously, predominantly in the ML [arrowheads in (E) and (G)]. For comparison, an 8-month-old noninjected male APP23 mouse (J and K) and a 5-month-old noninjected male APPPS1 mouse (L and M) are shown. Scale bar, 350 μm. (I) Urea-based SDS-PAGE immunoblot analysis with an antibody specific to human Aβ. Lane 1: APPPS1 extract; lane 2: APP23 extract. Aβ1-42 was the major Aβ species in APPPS1 mice, whereas Aβ1-40 predominated in the APP23 extract. Total Aβ was comparable in the two extracts [APPPS1: total Aβ, 14.7 ng/μl (Aβ40: 4.6 ng/μl; Aβ42: 10.1 ng/μl); and APP23: total Aβ, 11.7 ng/μl (Aβ40: 8.1 ng/μl; Aβ42: 3.6 ng/μl)].

To establish whether Aβ is a prerequisite for the amyloid-inducing activity of the extract, APP23 Tg extracts were either Aβ-immunodepleted (Fig. 3, A to C) or mixed with the Aβ-specific antibody (anti-Aβ) Beta-1 (Fig. 3, D to F) and injected into young APP23 hosts. Immunodepletion completely prevented the amyloid-inducing activity of the extract, whereas the Beta-1–containing extract attenuated amyloid induction by >60% compared to a control antibody.

Fig. 3.

(A to C) Immunodepletion: APP23 Tg brain (0.5%) extract (A) or the same extract Aβ-immunodepleted (B) was injected intrahippocampally into 3-month-old APP23 mice that were then analyzed 3 months later. Immunodepletion completely eliminated the amyloid-inducing activity of the extract. (C) Urea-based SDS-PAGE followed by immunoblotting with human Aβ-specific antibody. Lane 1: intact APP23 Tg extract; lane 2: immunodepleted extract (the absence of detectable Aβ and APP fragments is apparent); lanes 3 to 6: eluted pellet fractions after the first, second, third, and fourth depletion steps. (D to F) Antibody blocking: APP23 or AD brain extract (10%) was mixed with either a control antibody or with anti-Aβ Beta-1 and injected into 6-month-old male APP23 mice. Four months after injection, amyloid induction was significantly lower in the Beta-1–injected mice (E) than in the control mice (D), as confirmed by stereological quantification of immunoreactive Aβ (F) (n = 5 mice per group; P < 0.05). (G to I) Immunization: APP23 or AD brain extract (10%) was injected intrahippocampally into 6-month-old male APP23 mice, followed by weekly peripheral injections of Beta-1 antibody (H) or control antibody (G), starting 1 month after the brain-extract injections. Four months later, immunohistochemical (G and H) and stereological (I) analysis revealed a >90% inhibition of amyloid induction by passive immunization (n = 5 mice per group; P < 0.001). (J to L) Formic acid treatment: APP23 Tg extract (10%) was treated with formic acid and injected into the hippocampus of 3-month-old APP23 mice that were analyzed 3 months later. (K) Formic acid completely nullified the amyloid-inducing activity of the extract. (J) PBS-treated control extract. (L) Bicine-Tris SDS-PAGE (without urea) followed by immunoblot analysis with human Aβ-specific antibodies. Lane 1: synthetic Aβ (2 ng/μl); lane 2: PBS-treated control APP23 extract (10 μl); and lanes 3 and 4: formic acid–treated APP23 extract (20 μl in lane 3 and 10 μl in lane 4). The oligomeric bands are absent or greatly reduced (arrowheads) in the formic acid–treated extract, even at twice the concentration of the control extract. (M to O) Heating: APP23 extract (10%) was heated to 95°C (N) before injection into the hippocampus of 3-month-old APP23 mice. Analysis 3 months later showed that heating reduced (45%) but did not eliminate Aβ seeding (n = 3 mice per group; P > 0.05). Scale bar, 350 μm.

Passive immunization of APP23 host mice with Beta-1 antibodies inhibited the development of induced lesions. Administration of the antibodies commenced 4 weeks after the intracerebral injection of APP23 Tg extract in order not to interfere with the initial seeding process. Serum anti-Aβ titers of 1:2400 to 1:8000 were maintained until mice were killed 4 months after the extract infusion. Amyloid induction was almost completely inhibited in immunized mice compared to those injected with control antibody (Fig. 3, G to I). An active immunization protocol revealed similar inhibition (fig. S5).

The amyloid-inducing activity of the extract was disrupted by formic acid denaturation. APP23 Tg extracts were treated for 1 hour with 70% formic acid, followed by dialysis, or heated to 95°C for 5 min, cooled, and injected into young APP23 mice. Formic acid treatment completely abolished the amyloid-inducing activity of the extract, whereas heating reduced, but did not eliminate, amyloid induction (Fig. 3, J to O).

Preparations of soluble or fibrillar synthetic Aβ40, Aβ42, or a mixture of both, in amounts similar to those of Aβ in APP23 Tg brain extracts, failed to induce detectable Aβ deposition (Fig. 4). Injection of synthetic Aβ at concentrations 100 to 1000 times that of Aβ in APP23 Tg brain extracts resulted in amorphous masses of material near the injection site that consisted primarily of the injected material. Few newly generated aggregates were present, as confirmed by infusion of biotinylated Aβ1-42 (fig. S6). Synthetic Aβ oligomers (1416) also were ineffective at inducing parenchymal and vascular amyloid, as was Aβ isolated from the conditioned media of cells stably transfected with APP (table S1; fig. S6). Addition of cofactors such as ApoE4 and Cu/Zn, which are thought to promote the polymerization of Aβ (17, 18), did not augment the potency of synthetic material (table S1). Injection of synthetic Aβ mixed with brain extract from wild-type mice also did not produce notable β-amyloidosis (table S1). Finally, the possibility that poor seeding by synthetic Aβ results from the activation of the Aβ-degrading enzyme neprilysin (19) or a humoral immune response (20) was ruled out by unchanged neprilysin immunoreactivity and by the absence of serum anti-Aβ antibodies in mice injected with synthetic Aβ.

Fig. 4.

Fresh and aged Aβ preparations were made from Aβ1-40 or Aβ1-42, or from a 2:1 mixture of Aβ1-40 and Aβ1-42 at ∼5 ng/μl, similar to the Aβ concentrations in a 10% APP23 brain extract. (A) Urea-based SDS-PAGE followed by immunoblotting with a human Aβ-specific antibody. Lane 1: APP23 Tg brain extract; lane 2: mixed fresh Aβ1-40 + Aβ1-42; lane 3: fresh Aβ1-42; lane 4: fresh Aβ1-40; lane 5: aged Aβ1-40 + Aβ1-42; lane 6: aged Aβ1-42; and lane 7: aged Aβ1-40. Preparations were injected intrahippocampally into 5-month-old male APP23 mice that were analyzed 4 months later. The aged Aβ preparations were fibrillar in nature as verified by electron microscopy and Congo red binding (fig. S6). Multimeric Aβ species are not easily detected with this urea-based gel system, which is designed to separate the different Aβ isoforms. The additional band in the brain extract is likely Aβ1-38. (B) Amyloid induction with an APP23 Tg extract. (C and D) No induced Aβ deposits were detectable with any of the synthetic Aβ preparations at this concentration. Shown are animals injected with freshly mixed Aβ1-40 + Aβ1-42 (C) and aged Aβ1-40 + Aβ1-42 (D). Scale bar, 350 μm.

Thus, cerebral extracts induce Aβ deposition in vivo by supplementing (or anticipating) endogenously generated Aβ seeds with exogenous seeds that probably consist of a form of multimeric Aβ. The host-specific morphology and distribution of the induced lesions underscore the essential role of the host in regulating pathogenesis, but the inducing agent also contributes to the pathologic phenotype. The inhibition of seeding by specific immunoneutralization of Aβ (20) or formic acid denaturation of the extracts (21) suggests that the active agent consists of an aggregated Aβ species (20, 21). The finding that synthetic Aβ lacks amyloid-inducing activity in vivo was not unexpected, inasmuch as prion disease has also been difficult to transmit by in vitro–generated (recombinant) prions (8).

Synthetic and cell culture–derived Aβ, in concentrations similar to those tested in the present study, are neurotoxic in vivo (14, 22, 23) and can impair long-term potentiation and cognitive function (24, 25). These observations, in light of the highly variable seeding efficacy of in vitro and in vivo preparations, suggest the occurrence of various Aβ conformations with partially distinct biological activities (26), similar to prions (27, 28). Polymorphic and self-propagating synthetic Aβ strains recently have been reported (9). Thus, Aβ multimers in vivo also may be polymorphic and polyfunctional, again reminiscent of prions, in which infectivity is strain-dependent and fully encoded in distinct multidimensional conformations (29). Whether oligomeric forms of Aβ that are thought to be key cytoactive disease agents (25, 30, 31) can also be seeded in vivo remains to be determined.

There is currently no evidence that β-amyloidosis (and in particular AD) is transmissible in the same sense as are prion diseases, which can be transmitted to wild-type hosts via diverse routes of varying efficiency and involve systemic cellular mechanisms of prion uptake and distribution (7, 32). However, an understanding of the mechanisms involved in the instigation and propagation of abnormal Aβ assemblies in vivo could shed light on the origins of idiopathic Alzheimer's disease.

Supporting Online Material

Material and Methods

Fig. S1 to S6

Table S1


References and Notes

Stay Connected to Science

Navigate This Article