Injection of rAAV2 vectors expressing transgenes for human Aβ40, Aβ42, C100 and C100V717F into the mouse hippocampus and cerebellum resulted in wide-spread transduction in both brain regions and the development of some pathological changes characteristic of AD, most notably increased microgliosis and increased permeability of the blood brain barrier.
Injection of rAAV2-Aβ42-GFP and rAAV2-C100V717F
-GFP in the hippocampus resulted in significantly increased IBA-1 density at 3 months post-injection, while injection of rAAV2-Aβ40-GFP and rAAV2-C100-GFP did not, suggesting that Aβ42 may be a more potent mediator of microgliosis than Aβ40. It is known that Aβ attracts and causes activation of microglia 
, but it has been previously suggested that this may only occur in response to fibrillar Aβ 
. The association between fibrillar Aβ and microgliosis is further supported by the fact that the onset of gliosis in AD transgenic mice is closely linked to the onset of plaque deposition 
, and manipulating the amount of plaque deposition results in similar changes in the extent of microgliosis 
. However, the results from this study suggest that microgliosis can also occur in response to soluble forms of Aβ, as no plaques were observed after injection of any rAAV2 vector.
The rAAV2 vectors used in this study also resulted in pathology indicative of increased permeability of the blood brain barrier, which was most extensive in the hippocampus at 3 months post-injection. Injection of rAAV2 vectors into the hippocampus resulted in increased brain IgG and increased numbers of IgG/IBA-1 positive cells. IgG cannot cross the blood brain barrier and is only found in the brain under pathological conditions 
. As a result, IgG is a well established marker of blood brain barrier permeabilisation and has been shown to be a good alternative to other markers of blood brain barrier disruption such as Evans blue dye staining 
. Cells with similar morphology to the IgG positive cells observed in this study have been characterised previously and a large number of these cells is also a common marker of blood brain barrier disruption 
. Previous studies have suggested that these cells are leukocytes and as the cells observed in this study were also immuno-positive for IBA-1, this suggests that they were leukocytes of monocyte-macrophage lineage 
, in agreement with previous studies 
In the hippocampus at 3 months post-injection, blood brain barrier disruption was more extensive after exposure to Aβ42, via expression of Aβ42 directly or the C100 and C100V717F
precursors. In comparison, while IgG staining intensity and numbers of infiltrating cells after injection with rAAV2-Aβ40-GFP were elevated, these changes were not significantly different from that observed after injection with rAAV2-GFP. Blood brain barrier disruption is a pathological feature of AD 
and previous studies have hypothesised that it may be directly caused by Aβ 
. The results from this study not only support the hypothesis that Aβ expression may directly cause blood brain barrier disruption, but also suggest that Aβ42 may be a more potent mediator of blood brain barrier disruption than Aβ40.
Injection of rAAV2 vectors did not induce widespread astrogliosis or altered neuronal density in either the hippocampus or cerebellum. Activation of astrocytes in response to Aβ expression was observed to some extent, however it was primarily localised to the injection site, in contrast to the more extensive microgliosis. Previous studies have found Aβ to cause activation and migration of astrocytes 
. However, it has also been shown that the activation of astrocytes in AD is dependent upon the conformation and aggregation state of Aβ; astrocytes surrounding dense core plaques become activated, while astrocytes surrounding diffuse plaques or those not associated with aggregated Aβ do not show signs of activation and can often show signs of atrophy 
. Therefore, it is possible that the lack of extensive astrogliosis observed in this study was due to the lack of aggregated, fibrillar Aβ following injection of viral vectors. The lack of widespread neurodegeneration observed in this study is consistent with previous studies that have shown that Aβ is not a potent mediator of neurodegeneration in vivo
. It is now becoming more accepted that tau hyperphosphorylation and the development of neurofibrillary tangles is more likely to be a mediator of neurodegeneration in AD than Aβ 
. It is interesting that synaptophysin mRNA levels were unaffected as Aβ, particularly soluble oligomers of Aβ, has been shown to decrease expression of synaptic markers, including synaptophysin 
. However, this finding is not consistent across all studies 
, indicating that decreased synaptophysin expression may not be a direct consequence of Aβ expression and that other additional factors may be necessary. It is important to note that only gross neurodegeneration would have been observed by the quantification measures used in this study and that the use other neuronal and cell death markers may have provided a more specific indication of neurodegeneration.
Western blot and immunohistochemical processing failed to detect significant Aβ40 and Aβ42 protein expression in transduced brain regions. We do not believe that this was due to the use of inefficient viral vectors because Aβ42 was detected in tissues injected with rAAV2-C100V717F
-GFP or rAAV2-Aβ42-GFP using the more sensitive ELISA technique, thus proving that these vectors were capable of producing Aβ. Furthermore, there was strong transduction of all bi-cis
tronic rAAV2 vectors in the hippocampus and cerebellum, as shown by high expression levels of transgene mRNA and post-IRES GFP protein, the latter produced only after C100 or Aβ protein translation. Finally, injection of rAAV2-Aβ40-GFP and rAAV2-Aβ42-GFP resulted in the development of obvious brain pathology that was unique to each Aβ isoform and was not present after injection of vehicle or rAAV2-GFP controls, strongly suggesting that the Aβ produced as a result of rAAV2 vector injection was responsible for the observed pathologies. We hypothesise that the inability to detect Aβ using the less sensitive immunohistochemistry and western blotting methods, and the observation of variance between animals in the amount of Aβ detected using ELISA, was due to rapid clearance and/or degradation of Aβ. Aβ levels are highly regulated in vivo
by rapid clearance across the blood brain barrier, phagocytosis by glia and degradation by multiple enzymes. Enhancement of any of these mechanisms could result in low Aβ levels. Indeed, some of the pathology observed in vivo
including microgliosis and infiltration of cells from the periphery indicate that Aβ degradation may have been increased in response to expression of Aβ as both microglia and infiltrating monocytes are capable of Aβ phagocytosis and enhance degradation 
. Increased Aβ clearance may also explain the lack of Aβ40 detected using the ELISA-mutiplex assay, as this isoform is more readily cleared and degraded than Aβ42 
. Furthermore, the Indiana mutation promotes the preferential production of Aβ42 rather than Aβ40, therefore the presence of this mutation and the hypothesised increased clearance of Aβ could account for the very low levels of Aβ40 observed after injection of rAAV2-C100V717F
-GFP. Lack of available tissue prevented ELISA-based quantification of Aβ40 levels after injection of rAAV2-Aβ40-GFP.
An intriguing finding was that, while significant levels of Aβ42 protein were detected in the cerebellum but not in the hippocampus, injection of rAAV2 vectors into the hippocampus resulted in greater pathological changes than those seen in the cerebellum. Why there was this apparent paradox of increased Aβ42 but reduced pathology in the cerebellum is not clear, but these observations do suggest that there are differences in the way different brain regions process and respond to C100 and/or Aβ. Previous studies have tried to determine why the cerebellum is less vulnerable to AD pathology. It has been shown that the cerebellum contains all of the necessary proteins to produce Aβ 
, and that plaques do eventually appear in the cerebellum as AD pathology advances 
, indicating that the cerebellum is capable of producing amyloid pathology. Nonetheless, the cerebellum consistently has fewer plaques and lower levels of insoluble Aβ and intracellular Aβ42 
than other brain regions that are primarily affected in AD such as the hippocampus and cortex. It seems that the cerebellum is better equipped to prevent AD pathology from progressing. A recent study reported that secreted metabolites produced from cerebellar neurons reversed AD brain pathology in AD transgenic mice, while metabolites from hippocampal neurons exacerbated pathology 
. The exact proteins or pathways involved in the protection of the cerebellum in AD are not yet known, but it has been suggested that this may be specifically due to enhanced clearance or degradation of Aβ 
. The present data suggest an alternative hypothesis, that cerebellar cells may be intrinsically less responsive to the presence of Aβ and/or C100. Future research is needed to further examine why AD brain pathology develops differently in different brain regions as this could help determine what initiates the development of AD brain pathology.
Vectors expressing the C100 transgene were more effective at consistently producing higher amounts of Aβ than vectors directly expressing Aβ transgenes, both in vitro
and in vivo
. This most likely resulted from the more physiological method of production of Aβ from C100, in comparison to the non-physiological production by direct expression of either Aβ40 or Aβ42. Direct expression of Aβ may not be optimal for Aβ accumulation, possibly due to Aβ production occurring in the incorrect sub-cellular location. Previous in vitro
studies have shown that fusing Aβ and C100 to a signal protein that directs expression in the secretory pathway greatly increases the amount of Aβ detected after plasmid transfection 
, hence suggesting that sub-cellular location of Aβ may be important for expression.
A further aim of this study was to determine if the effects of transduction with rAAV2 vectors expressing APP fragments were exacerbated at 6 months post-injection in comparison to 3 months post-injection. This was not found to be the case in either brain region as less extensive pathological changes were observed at 6 months post-injection. The level of transduction was similar at 3 and 6 months post-injection, therefore the less extensive pathology observed at 6 months post-injection is unlikely to be a result of any technical issues associated with long-term transduction. Instead, it is possible that brain regions may have adapted to the long-term expression of C100 and/or Aβ and as a result became better equipped to deal with the consequent pathology, such as by increasing levels of Aβ degrading enzymes or increasing anti-inflammatory proteins. However, further studies are necessary to confirm this hypothesis.
In conclusion, the use of viral vectors to over-express Aβ and C100 is a promising technique with which to examine the consequences of Aβ expression in mature CNS tissues in vivo. Results from this study provide evidence that Aβ42 causes greater pathology than Aβ40, particularly by promoting microgliosis and inducing abnormal permeability changes in the blood brain barrier.