Here we show that a transient reduction of Aβ*56 corresponded with the full restoration of spatial reference memory in Tg2576 mice. The data indicate that despite an increasing amount of amyloid plaque in the brain, the reduction or removal of Aβ*56 was associated with improved memory function. Taken together with previous results showing impaired memory in mice lacking amyloid plaques but producing Aβ*56 (Lesne et al., 2006
), we conclude that in the early stages of amyloid deposition there is no association between plaques and memory loss, in contrast to a strong association between Aβ*56 and memory loss.
The data suggest that a transient change in the kinetics of plaque formation takes place at ~12 months of age, which drives Aβ toward fibrillar Aβ formation and results in a dramatic reduction in Aβ*56 levels. Although we cannot exclude the alternative possibilities that soluble Aβ assemblies are cleared or degraded during this specific time interval, several reasons make this unlikely: i) the transient reduction in Aβ oligomers is extremely brief (only a few weeks); and ii) Aβ*56 is resistant to harsh conditions ex vivo
. 4% SDS, 10% HFIP) (Lesne et al., 2006
). Moreover, although the susceptibility of Aβ*56 to proteolytic catabolism in vivo
is unknown, it is unlikely that there is a transient proteolytic process occurring for a few weeks around 12 months of age that degrades oligomers but leaves monomers and plaques intact.
Considerable advances in our understanding of the neuronal effects of naturally derived Aβ oligomers, generated in cultured cells or in brain tissue, have been made in recent years. Although direct comparisons of in vitro- (cell lines, primary cultured neurons) and in vivo- (brain tissues) derived Aβ oligomers have not been done, the current literature suggests that they exert distinct effects. Cells transfected with variant APP genes may produce low-n
Aβ oligomers such as Aβ dimers and trimers. These low-n
oligomers reduce dendritic spine density in vitro
(Shankar et al., 2007
), and impair the maintenance, but not the initiation, of LTP (Walsh et al., 2002
). Thus, the data on cell-derived Aβ oligomers suggest that they are synaptotoxic.
In contrast, several lines of evidence indicate that brain-derived low-n
oligomers such as Aβ trimers in Tg2576 mice are not synaptotoxic. First, Tg2576 animals produce a constant level of Aβ trimers throughout life (Lesne et al., 2006
), but mice younger than ~10 months of age, prior to the formation of mature amyloid plaques, show no synaptic alterations (Spires et al., 2005
). However, synaptotoxicity is evident in older, plaque-bearing Tg2576 mice, with reductions in dendritic spine density occurring within a 100-micron halo surrounding the surface of plaques (Spires et al., 2005
). Second, evoked synaptic responses in 8 to 10 month-old Tg2576 mice are normal, but become abnormal at 14 months, an age when plaque deposition is substantial (Stern et al., 2004
). Third, the levels of Aβ trimers in 6 month-old Tg2576 mice do not correlate with memory dysfunction (Lesne et al., 2006
). These lines of evidence strongly argue against a synaptotoxic effect of the soluble Aβ trimers we measured in Tg2576 mice. Moreover, they lessen the possibility that in the current study the improvement in memory function in ~12-month Tg2576 mice was due to the lowering of the Aβ trimers that were reduced along with Aβ*56.
The published studies of naturally produced Aβ oligomers indicate that there may be two distinct forms of low-n Aβ oligomers in the Tg2576 brain. Clearly, there are non-synaptotoxic Aβ trimers that are not intimately associated with amyloid plaques. Still unclear is the nature of the synaptotoxic Aβ species, or other agents, located within the 100-micron wide halo surrounding the amyloid plaques. However, they could potentially resemble cell-derived low-n Aβ oligomers, given their identical damaging effects on dendritic spines.
Tg2576 mice lack neurodegeneration (Irizarry et al., 1997
), and may therefore be a model of early or pre-clinical AD (Ashe, 2001
, Lesne et al., 2006
). Our results suggest that strategies aimed at reducing Aβ*56 could be an effective approach for treating patients suffering from early AD-related memory problems. Conversely, strategies that reduce amyloid burden without concomitantly decreasing Aβ*56 are less likely to be effective.