DISCUSSION

Our study examined whether there might be subtle neuronal loss in a transgenic mouse model of AD with robust $A\beta$ deposition, particularly $A\beta$ deposit associated neuronal loss. Prior stereologic based studies in several APP transgenic model systems demonstrated barely detectable or no neuronal loss, which contrasts with human AD where 50of hippocampal CA1 or entorhinal neurons are lost despite $A\beta$ burdens $1/2$ to $1/3$ less than those of the 12 month PSAPP mice. Since these transgenic mice do not develop neurofibrillary tangles, the PSAPP mice (and other mouse models) can determine the effects of $A\beta$ deposition in its various forms on the brain. Focal changes associated with $A\beta$ deposits in transgenic mice include astrocytosis, microgliosis, and neuritic dystrophy. To investigate focal neuronal loss associated with $A\beta$ deposits, we determined the radial density function associated with each $A\beta$ deposit. Based on the form of the radial density function, we find that: (i) larger and denser $A\beta$ immunoreactive deposits are associated with neuronal loss, (ii) the $A\beta$ deposits associated with neuronal loss are almost exclusively ThioS positive, and (iii) there is no penumbra of toxicity beyond the ThioS staining deposit. Modeling the $A\beta$ deposit interaction with neurons as a biophysical many-body interaction problem, computer simulations indicate that ThioS staining deposit behave as toxic, space-occupying lesions.

These results are consistent with in vitro studies, where aggregated $A\beta$ is more toxic to neurons in culture than solubilized monomeric A=DF [13,15]. ETC., ETC.

Previous stereological analyses, including our own, have not detected substantial neuronal loss in APP or PSAPP transgenic mice, whereas our current analysis is providing strong evidence for $A\beta$ associated neurotoxicity. This apparent discrepancy is easily resolved, however. The current analysis shows that the vast majority of $A\beta$ deposits do not alter the neuronal landscape; in fact, only the densest and largest $A\beta$ deposits (less than $3$% area coverage) are associated with neurotoxicity. For example, ThioS cores occupy only 1.5% of the cortical surface. Even if all neurons within ThioS cores were lost, this difference in total neuron numbers would be difficult to detect. As an example, given animal to animal variation of 5-10%, we calculate that 175 animals would need to be counted to detect 1.5% loss. In other words, focusing on the microenvironment around individual $A\beta$ deposits substantially increases the power of our analysis to detect neurotoxicity. Note change from Thios staining deposits to Thios core

In conclusion, the neurotoxic effects of $A\beta$ in the brain of PSAPP mice are limited to the largest and densest deposits that have a $\beta$-pleated sheet conformation. Although a small percentage of all $A\beta$ deposits, these $A\beta$ deposits have a marked effect on neurons in their immediate environment. The results suggest that $A\beta$, when in a $\beta$-pleated sheet conformation, is toxic to cortical neurons in PSAPP mice, while non-fibrillar $A\beta$ does not kill neurons. This result is analogous to observations in cell culture, in which fibrillar $A\beta$, but not soluble $A\beta$, is toxic; the current results are also consistent with the idea that the $\beta$-pleated sheet conformation of $A\beta$ is important for toxicity in vivo [35], perhaps by inducing cross linking or conformational changes of cell surface proteins [35]. Taken together, our data show that fibrillar $A\beta$ toxicity is one of the mechanisms of neuronal death active in APP transgenic mice, and, by extension, in AD itself.

The in vivo results are consistent with in vitro data suggesting a toxic gain-of-function of A$\beta$ when it adopts a densely packed, $\beta$-pleated sheet conformation [35].