INTRODUCTION

The primary pathologic features of Alzheimer's disease (AD) brain are amyloid deposition, neurofibrillary tangle formation, and neuronal loss. There is substantial evidence implicating amyloid beta protein ($A\beta$) in the pathological cascade leading to neuronal loss in AD [1]. Presenilin-1 (PS1), presenilin-2, and amyloid precursor protein (APP) mutations causing familial AD and the apoE e4 allele risk factor for AD all increase increased plasma, fibroblast or brain levels of $A\beta$ or $A\beta$x-42/43 in AD and transgenic mice [2,3,4,5,6,7,8,9,10]. Results of studies linking $A\beta$ with neuronal death, however, have been paradoxical, and are influenced by the aggregation state of $A\beta$. $A\beta$ neurotoxicity has been clearly demonstrated in cell culture [11,12,13,14,15,16,17]. However, in human AD, the total amount of extracellular $A\beta$ has little or no correlation with the amount of neuronal loss in the human AD brain, which exceeds 50% in vulnerable regions like the hippocampal formation, entorhinal cortex, and association cortex [18,19,20].

The availability of several transgenic models of AD that develop cerebral $A\beta$ deposition has allowed the in vivo assessment of the pathological effects of $A\beta$ deposition [8,21,22,23,24,25,26]. In all models, $A\beta$ deposition is associated with focal changes of inflammation (astrocytosis and microgliosis), neuritic dystrophy, and tau phosphorylation. Quantitative stereological studies of neuron number in various brain regions in PDAPP (APPV717F), Tg2576 (APPSw), PSAPP (APPSw x PS1M146L), and APP23 mice demonstrate either no or minimal loss of neurons [27,28,29,30]. Nonetheless, qualitative observation suggests that some $A\beta$ deposits in the transgenic mice do seem to alter the integrity of the neuronal architecture (Fig. 1(a)).

In order to reconcile cell culture studies showing marked $A\beta$ toxicity with studies of APP transgenic mice showing preservation of neurons despite extraordinary levels of $A\beta$ deposition, we hypothesized that only a subset of $A\beta$ deposits is biologically toxic in transgenic mice. Here we develop a new technique to assess neuronal integrity in each $A\beta$ deposit's microenvironment-that is, the local neuronal density within and surrounding each $A\beta$ deposit. We find strong evidence that a subset of $A\beta$ deposits-specifically compact, fibrillar ThioS deposits-are associated with neuronal loss in PSAPP transgenic mice. Furthermore, only the fibrillar core of $A\beta$ deposits appears to be toxic, and not the associated non-fibrillar surrounding $A\beta$ penumbra. Mathematical modeling of the effect of ThioS positive $A\beta$ deposits on adjacent neurons is compatible with the idea that these deposits are toxic to neurons, rather than acting as a non-toxic mass lesion pushing neurons away.