To better define the contribution of individual Aβ peptides to the cognitive deficits and amyloid deposition observed in models of AD, virally-mediated gene transfer of BRI-Aβ fusion proteins was used to increase hippocampal levels of individual Aβ species. Use of BRI-Aβ fusions results in enhanced Aβ secretion in the absence of APP overexpression, and distinguishes this approach from overexpression of Aβ minigenes, a strategy that generates high levels of intracellular Aβ but minimal secreted Aβ [30
]. Animals injected with AAV1 vectors encoding APPsw, BRI-Aβ40 or BRI-Aβ42, alone and in combination, developed behavioral deficits in a distinct pattern. AAV-APPsw and AAV-BRI-Aβ42 animals had reduced exploration behavior during working memory evaluation but no significant deficits in passive avoidance or acquisition and retention of spatial information. Animals injected with AAV-BRI-Aβ40 alone were impaired in passive avoidance, but were comparable to age-matched naive controls in the working memory task and were not significantly impaired in the Morris water maze. However, animals co-injected with both BRI-Aβ vectors showed the most pronounced behavioral deficits with some impairment in all tests. Despite measurable impairments occurring in all groups, only BRI-Aβ42 animals developed extracellular Aβ deposits. Taken together with the ELISA and histology results, this behavioral data confirms observations from AD transgenic mouse models that measurable behavioral deficits are not dependent on the presence of Aβ plaques. Unexpectedly, the data shows development of more pronounced cognitive deficits when Aβ42 and Aβ40 are co-expressed, and suggests a role for Aβ40, along with Aβ42, in cognitive impairment.
In the current study only AAV-BRI-Aβ42 animals developed extracellular Aβ deposits, located within the hippocampus. These diffuse deposits were immuno-positive for Aβ, but did not stain with thioflavin S or Congo Red and were not associated with astrogliosis or proliferation of microglia, indicating they are "non-cored" diffuse structures. Expression for 9 months enhanced Aβ42 accumulation but still did not result in formation of cored plaques (not shown). Thus, these structures are similar to the diffuse Aβ deposits observed in humans that are primarily composed of Aβ42 and not associated with significant reactive pathology. It is not clear why AAV-BRI-Aβ42 rats do not develop cored plaques. BRI-Aβ42 transgenic mice develop cored plaques in the cerebellum as early as 3 months of age [4
], and both diffuse and cored plaques in the forebrain by 12 months of age. However, in those animals brain levels of insoluble Aβ42 were markedly higher than the levels achieved in the current study. In contrast Tg2576 (APP695
SWE) mice, which have a high ratio of Aβ40 to Aβ42, predominantly develop cored plaques [7
]. Brain-region or species specific factors might regulate Aβ aggregation into diffuse or cored: in humans certain regions of the brain seem more prone to develop diffuse deposits of Aβ, and other factors such as ApoE [32
] and complement [36
] may also play a role in driving amyloid formation. In any case, the current study shows that viral delivery of BRI-Aβ42 can foster considerable Aβ accumulation in a relatively short time-frame, and confirms both transgenic Aβ Drosophila
] and transgenic BRI-Aβ mice studies [4
] where visible Aβ deposits were obtained only with Aβ42, but not Aβ40, overexpression.
The lack of correlation between presence of plaques and severity of cognitive dysfunction observed in the current study has been noted in a number of AD models [9
]. The current results are consistent with studies in transgenic mice that demonstrate behavioral effects are not correlated with visible plaques, but may correlate better with other Aβ assemblies. Indeed, in Tg2576 mice the appearance of behavioral deficits is associated with the initial occurrence of insoluble Aβ accumulation at a time when no overt plaque formation is noted [9
]. In the current study, overexpression of APPsw resulted in a pattern of deficits similar to that observed in the BRI-Aβ42 group but with no deposition of plaques or detectable increase in insoluble Aβ42, also consistent with data that changes in morphologic markers of synaptic integrity such as dendritic spine density and onset of behavioral deficits can precede a measurable rise in insoluble Aβ42 levels [48
Data obtained from the combined vector animals suggest some interplay between Aβ40 and Aβ42 levels resulting in enhanced behavioral deficits when the two peptides are co-expressed. The level of insoluble Aβ40 is higher in the BRI-Aβ40+42 animals than in those injected with AAV-BRI-Aβ40 alone. Despite an absence of Aβ deposits that are visible by immunohistochemistry, the presence of SDS-insoluble, formic acid-soluble Aβ indicates that the combined expression of Aβ40 and Aβ42 peptides does lead to insoluble Aβ accumulation, perhaps indicating seeding of Aβ40 deposition by Aβ42. Over-expressing both Aβ40 and Aβ42 in the absence of other APP fragments could result in production of a transient assembly, or a structurally distinct aggregate, that affects behavior to a greater degree than either peptide alone, as seen in a recent study where cognitive dysfunction in Tg2576 mice was linked to formation of a transient soluble assembly [49
]. The biochemical analysis of Aβ species described in the current study were conducted prior to these recent reports; thus, the material has been extracted in a manner that would prevent analysis of Aβ * and other oligomeric Aβ species. The behavioral impairments observed in the combined vector treatment group could also be explained by increased anxiety and locomotor behavior – in the open field these animals crossed significantly more lines and spent less time in the open than all other groups. Increased anxiety itself could be indicative of altered hippocampal functioning or damage [50
The exact role of Aβ40 in cognitive impairment is currently unclear. In human patients high plasma concentrations of Aβ40 are associated with an increased risk of dementia [54
] and along with Aβ42, Aβ40 is known to impair hippocampal LTP in rats [55
] and to alter glutamate receptor composition and trafficking and synaptic function [58
]. However, studies to date on specific effects of Aβ on cognitive function in mammalian brain have relied on non-specific pharmacological intervention to increase Aβ levels [60
] or infusion of Aβ peptides directly into the hippocampus or ventricles [61
] rather than the prolonged and more physiologic secretion strategy adopted here. The pathology of transgenic BRI-Aβ40 and BRI-Aβ42 mice has been characterized [4
], but the behavioral phenotype of these animals has not been reported, and to date no behavioral phenotype has been detected (E. McGowan, T. Golde, C. Janus personal communication). A recent study of transgenic Drosophila
supported a role for both Aβ40 and Aβ42 in age-dependent learning defects [38
], with a higher level of Aβ40 than Aβ42 required to affect learning ability. Our data confirm a potential role for both Aβ40 and Aβ42 in altering cognitive impairment that, at least for Aβ40, appears to be dissociable from overt plaque formation. The effects of Aβ40 and Aβ42 in AAV-BRI-Aβ treated rats may reflect alterations in glutamate receptors and synaptic assemblies, and future work will include examination of synaptic markers, NMDA receptor composition and PSD95 levels.