Soluble Aβ oligomers are thought to cause early synaptic damage and memory deficits in AD , although the mechanisms through which Aβ aggregates might lead to this phenotype are not yet fully understood. During the progression of AD, monomers of Aβ can aggregate to form amyloid fibrils. Five distinct fibrillar aggregates induced by Zn2+ have been described , including protofibrils, Aβ-derived diffusible ligands (ADDL) and oligomeric species [34, 35]. Oligomeric Aβ peptides have the ability to form dimers, trimers, tetramers and higher-order arrays that can form so-called annular structures. These are thought to influence the functionality of cytoskeleton-associated proteins, cause damage to synaptic spines and inhibit long-term potentiation in cultured neurons [33, 36, 37] and in vivo . It was shown that physiological levels of Cu2+ and Zn2+ cause Aβ to aggregate  and that Zn2+ ions are bound to Aβ via the histidine imidazole rings within senile plaque cores . Recently, Adlard et al. proposed a mechanism whereby Aβ pathology causes cognitive impairment by trapping synaptic Zn2+ rather than through direct toxicity . Thus, the transsynaptic movement of Zn2+ may be severely compromised in AD by being sequestered in Aβ. This trapping of Zn2+ might mimic ZnT3 ablation  and indeed, mice with a disruption of the vesicular Zn2+ transporter ZnT3, display complete absence of Zn2+ from synaptic vesicles throughout the brain  as well as synaptic and memory deficits comparable to those seen in a cognitively impaired APP transgenic mouse model of AD .
Based on these findings, we propose a model, where Zn2+-ions might fail to reach their postsynaptic targets like ProSAP/Shank proteins due to sequestration by Aβ, leading to a dysregulation of the PSD scaffold and ultimately to a loss of synapses that can also be seen in ProSAP/Shank knockdown conditions . This model is consistent with findings of Deshpande et al., who postulated that sequestration of Zn2+ in oligomeric Aβ leads to reduced availability of Zn2+ at the synapse, ultimately leading to cognitive deficits in AD . To test this model, we investigated the influence of Aβ1-40 and Aβ1-42 on ProSAP/Shank family members in hippocampal neuron culture. In line with a number of recent publications showing the possibility that Aβ oligomers influence synaptic proteins and thus interfere with synaptic function [7, 12, 13, 44, 45], our study shows that the synaptic levels of ProSAP2/Shank3 and Shank1 decrease significantly following the addition of Aβ to primary neurons. Moreover, introduction of Aβ oligomers leads to a significant reduction (about 25%) in synapse density in hippocampal cultures, which is in agreement with previous studies reporting 11 to 77% declines in synaptophysin immunostaining in brain sections [46, 47]. These results are also consistent with recent studies in cellular and rodent models, showing that small soluble oligomers are toxic because they directly damage synapses [32, 48]. Furthermore, our experiments show that the loss of synapses is caused by a decrease in mature synapses. Thus, we conclude that the reduction in synapse density caused by Aβ is due to impaired activity dependent maturation and destabilization of mature synapses, but leaves the ability of an initial formation of synapses intact.
Additionally, treatment of hippocampal neurons with Aβ1-40 leads to a significant downregulation of ProSAP2/Shank3 at the synapse, to an impairment in synapse maturation and, in line with previous studies, to a downregulation of synaptic Shank1 levels . The decrease in synaptic ProSAP2/Shank3 is also reflected by a decrease in protein levels in the P2 fraction as assessed by Western Blotting after 24 h treatment with Aβ1-40. Given the multiple interaction partners of ProSAP/Shank proteins at the synapse, it is likely that Aβ mediated changes in ProSAP/Shank complex formation cause synaptic dysfunction induced by reducing actin cytoskeletal assembly, spine motility as well as the maturation and plasticity of excitatory glutamatergic synapses.
We also show that the observed changes in ProSAP/Shank levels at the synapse are not due to altered gene expression, proteasomal degradation or protein synthesis and it appears that other posttranscriptional mechanisms control synaptic ProSAP/Shank levels. One interesting candidate is Zn2+, which is known to bind and regulate the synaptic localization of specific ProSAP/Shank family members, including ProSAP1/Shank2 and ProSAP2/Shank3 but not Shank1 [16–18]. We thus investigated whether an increased demand on extracellular Zn2+, e.g. by an increased level of Aβ, would reduce cellular levels of Zn2+ and consecutively the synaptic levels of ProSAP/Shank family members. Using a cell-based assay, we directly demonstrated that the presence of extracellular Aβ interferes with the proper loading of ProSAP2/Shank3 with Zn2+. In contrast, saturation of Aβ with Zn2+ before application does not change ProSAP2/Shank3 Zn2+ loading.
In hippocampal cell culture, exogenously applied Aβ clusters with Zn2+ intracellular and treatment of cultured neurons with Aβ reduces dendritic Zn2+ levels. It was demonstrated previously that some intracellular Aβ is derived from extracellular Aβ pools and several distinct pathways of entry for extracellular Aβ have been proposed [49, 50]. Although intracellular accumulation of Aβ is seen in multivesicular bodies and lysosomes, it can also be found within the cytosol . Indeed, Kandimilla et al. have shown that Aβ is internalized by neurons primarily via passive diffusion . That way, a fraction of intracellular accumulating Aβ might directly compete with Zn2+ binding proteins such as ProSAP2/Shank3 for Zn2+ ions in addition to the sequestration of extracellular Zn2+ ions.
Based on these findings, we predicted that supplementation of hippocampal cultures with Zn2+ during the treatment with Aβ or application of Zn2+-saturated Aβ would lead to a rescue of the observed loss-of-ProSAP2/Shank3 phenotype. Our results show that the Aβ-induced decrease in synapse density as well as lowered synaptic levels of ProSAP2/Shank3 can indeed be rescued by Zn2+-supplementation. Moreover, Zn2+ saturated Aβ causes significantly less changes in synapse density and ProSAP2/Shank3 levels. Interestingly, also the decrease of Shank1 that shows a stronger requirement of NMDAR activity compared to ProSAP2/Shank3, can be rescued by Zn2+-supplementation. This indicates that Shank1 scaffold plasticity might depend on both, homeostatic changes via ProSAP2/Shank3 and the presence of Zn2+ ions as well as on changes induced by synaptic activity, driven by the activation of downstream signaling pathways.
Our findings are further supported by in situ studies using APP-PS1 mice and AD patient brain sections. Here, we observed that Zn2+ ions are enriched within amyloid plaques present in the hippocampus of older APP-PS1 mice and patients with severe AD. Intriguingly, intracellular Zn2+ concentrations are ~20% lower in neurons from these sections compared to control sections. However, in addition to the sequestration of Zn2+ by Aβ, other mechanisms may contribute to decreased intracellular Zn2+ concentrations, for example Metallothioneins (MTs) or other Zn2+-binding proteins such as α2 macroglobulin (A2M)  may alter levels by regulating intracellular free Zn2+. MT upregulation, as reported for MT-I in AD mouse models , leads to inhibition of NO-mediated Zn2+ release. Furthermore, pro-inflammatory cytokines cause a large induction of MTs . Several Zn2+ transporter proteins, including ZnT-1, ZnT-4 and ZnT-6, are altered in brain regions of subjects with early and late stages of AD . Moreover, several members of the ZnT family (ZnT-1, 3, 4, 5, 6, 7) are expressed in amyloid plaques .
In addition to reduced intracellular Zn2+ levels, we found a significant decrease in synapse density and synaptic ProSAP2/Shank3 and Shank1 protein levels. While chelation of Zn2+ by extracellular Aβ appears a likely mechanism for influencing Zn2+ levels in the brain, it should be noted that intracellular chelation of Zn2+ might also contribute to its sequestration. Interestingly, it was recently found that serum Zn2+ concentrations were significantly reduced from 12.3 μmol/l to 10.9 μmol/l in AD patients compared to control subjects . Moreover, Zn2+ supplementation greatly delays hippocampus-dependent memory deficits and strongly reduces both Aβ and tau pathology in the hippocampus of an AD mouse model .
However, distinct mechanisms might contribute to the observed decreases in PSD scaffold proteins in a brain region specific manner. In cortical cultures, the Aβ1-40-mediated reduction of PSD-95 protein levels is dependent on NMDAR activity and cyclin-dependent kinase 5, involving the proteasomal pathway . However, the decreased levels of Homer1b and Shank1 were not influenced by proteasome activity. The decreased levels of synaptic Homer1b required de novo protein synthesis and involved the PI3-K pathway and calcineurin phosphatase (PP2B) activity, whereas declustering of Shank1 required NMDAR activity and activation of the ERK pathway . In this study, the focus on the hippocampal region and the use of primary cultured neurons derived from hippocampus might explain the differences in regulatory pathways and kinetics mediating decreased levels of PSD scaffold proteins. This is underlined by our results, showing that a downregulation of ProSAP2/Shank3 and Shank1 in cortical neuronal cultures indeed occurs already after 1 h treatment with Aβ as reported previously . Given that the hippocampus is the brain region with the highest Zn2+ concentration, Zn2+-dependent regulatory mechanisms of PSD plasticity might be more pronounced in the hippocampus compared to other brain regions.
Although sporadic forms of AD are the most common, mutations in presenilin are associated with familial AD causing approximately 50% of these cases. In fact, it was recently reported that presenilin is important for cellular copper and zinc turnover, having the potential to affect Aβ aggregation indirectly through metal ion clearance . Moreover, inflammatory processes that have been associated with AD  lead to a dysregulation of metallothioneins that might additionally sequester Zn2+. Thus, our experiments provide additional evidence for a common mechanism of the pathology of AD caused by the dysregulation of Zn2+ levels within the brain.