Pathology Associated with AAV Mediated Expression of Beta Amyloid or C100 in Adult Mouse Hippocampus and Cerebellum

Accumulation of beta amyloid (Aβ) in the brain is a primary feature of Alzheimer’s disease (AD) but the exact molecular mechanisms by which Aβ exerts its toxic actions are not yet entirely clear. We documented pathological changes 3 and 6 months after localised injection of recombinant, bi-cistronic adeno-associated viral vectors (rAAV2) expressing human Aβ40-GFP, Aβ42-GFP, C100-GFP or C100V717F-GFP into the hippocampus and cerebellum of 8 week old male mice. Injection of all rAAV2 vectors resulted in wide-spread transduction within the hippocampus and cerebellum, as shown by expression of transgene mRNA and GFP protein. Despite the lack of accumulation of Aβ protein after injection with AAV vectors, injection of rAAV2-Aβ42-GFP and rAAV2- C100V717F-GFP into the hippocampus resulted in significantly increased microgliosis and altered permeability of the blood brain barrier, the latter revealed by high levels of immunoglobulin G (IgG) around the injection site and the presence of IgG positive cells. In comparison, injection of rAAV2-Aβ40-GFP and rAAV2-C100-GFP into the hippocampus resulted in substantially less neuropathology. Injection of rAAV2 vectors into the cerebellum resulted in similar types of pathological changes, but to a lesser degree. The use of viral vectors to express different types of Aβ and C100 is a powerful technique with which to examine the direct in vivo consequences of Aβ expression in different regions of the mature nervous system and will allow experimentation and analysis of pathological AD-like changes in a broader range of species other than mouse.


Introduction
There is a large body of evidence suggesting that beta amyloid (Ab) accumulation in the brain may be a primary cause of Alzheimer's disease (AD). However, exactly how Ab contributes to AD pathology and the mechanisms involved are still not yet fully understood. The two main isoforms of Ab are Ab40 and Ab42. While it is widely accepted that Ab42 is more neurotoxic than Ab40, surprisingly few studies have attempted to determine the individual physiological and pathological effects directly attributed to Ab40 and Ab42 in vivo and even fewer studies have directly compared the effects of Ab40 and Ab42 expression [1,2]. It is important to understand the physiological and pathological functions of each Ab isoform as many proposed therapies for AD act by modulating Ab levels [3,4,5,6].
It is difficult to determine the exact role of Ab in AD using the current available animal models, which are predominantly transgenic mice. Therefore it was the aim of this study to develop an alternative animal model for AD research using viral vectors to initiate localised expression of human Ab. There are many benefits associated with using viral vectors to develop animal models of disease. For example, viral vectors can rapidly express transgene proteins localised to desired brain regions, or even specific cell types. Viral vectors can also induce transgene expression in many species and at specific ages, hence preventing developmental or other unwanted compensatory variables in response to life-long transgene expression. Viral vectors also allow for the expression of multiple genes with much greater ease than in transgenic mice, a feature that is particularly important when studying a multifactorial disease such as AD.
To date, two studies have used viral vectors to directly express Ab. The first study developed viral vectors that produced Ab40-BRI or Ab42-BRI fusion proteins [2]. The BRI protein is involved in amyloid deposition in Familial British Dementia and Familial Danish Dementia, and fusion of BRI to Ab results in enhanced secretion of Ab-BRI fusion proteins. Expression of Ab42-BRI resulted in development of amyloid plaques at 3 months postinjection and injection of a combination of BRI-Ab42 and BRI-Ab40 vectors resulted in cognitive impairment. The second study developed a LV-Ab42 vector, which was injected into the primary motor cortex of rats [7]. While resulting brain pathology was only examined at the early post-injection time point of 4 weeks, Ab42 expression resulted in astrogliosis, increased TNF-a expression and increased tau phosphorylation.
In the present study, recombinant bi-cistronic adeno-associated viral vectors (rAAV2) were constructed, expressing either human Ab40, Ab42, C100 or C100 V717F . Direct expression of human Ab40 and Ab42 was performed in an attempt to determine the individual functions of each of these forms of Ab, while bypassing the possible confounding functional effects of other APP fragments such as sAPPa, N-APP, C100 and AICD, which are also produced during APP processing [8,9,10,11,12]. Viral vectors expressing C100, the immediate precursor to Ab, with and without the Indiana mutation (V717F) were also developed in case Ab expression and/or functional effects were influenced by production from a precursor protein. The Indiana mutation results in preferential production of Ab42 rather than Ab40 [13,14], so this strategy allowed for the indirect comparison of Ab40 and Ab42 by comparing the downstream effects of C100 and C100 V717F expression. Each of these rAAV2 vectors was injected into the adult mouse hippocampus and cerebellum, brain regions that are vulnerable or relatively spared in AD respectively. rAAV2-GFP and PBS injections served as controls. A targeted number of changes associated with increased Ab were examined, including glial reactivity (both astrogliosis and microgliosis), neuronal degeneration, blood brain barrier disruption and expression of apolipoprotein E (apoE), clusterin, synaptophysin and presenilin 1 (PS1).

Ethics Statement
All animal experimentation was approved by the UWA Animal Ethics Committee and conformed to published NHMRC guidelines.
Plasmid and rAAV2 Construction rAAV2 vectors were generated from the pTRUF12 plasmid, which contains two inverted terminal repeat sequences from AAV serotype 2, the inserted transgene under the control of the cytomegalovirus early enhancer-chicken beta actin (CMV-CAG) promoter and an additional intron sequence with an enhancer element. Four novel plasmids were generated containing transgene sequences for Ab40, Ab42, C100 and C100 V717F . The pTRUF12 plasmid without an inserted transgene was used as the GFP control plasmid. An internal ribosome entry site (IRES) allowed the bicistronic expression of the inserted transgene with enhanced GFP (eGFP), resulting in transgene and GFP production as individual proteins within each transduced cell. All plasmids were sequenced to ensure that the transgene sequences were in the correct orientation and that no unwanted mutations were present. rAAV2 vectors were produced using a previously described method [15]. An infection assay of HEK 293T cells was performed to ensure infectivity and titres were determined by quantitative PCR (qPCR) for viral genomic copies (GC) extracted from DNasetreated viral particles. Titres ranged from 1.4610 12 to 4.1610 12 GC/ml.

In vitro Transfection
HEK 293T, PC12 (A.T.C.C. CRL-1721) and HeLa cells were grown to approximately 90% confluency. HEK 293T and HeLa cells were maintained in DMEM media (Sigma) containing 10% fetal calf serum, 2 mM L-glutamine (Invitrogen), 100 Units/ml penicillin (Invitrogen) and 100 mg/ml streptomycin (Invitrogen). PC12 cells were maintained in DMEM media containing 5% fetal calf serum, 10% horse serum, 2 mM L-glutamine, 100 Units/ml penicillin, 100 mg/ml streptomycin and 0.1 mM non-essential amino acids (Invitrogen). All cell types were transfected using 4 mg of plasmid DNA and 10 ml of Lipofectamine 2000 (Invitrogen) according to the manufacturer's instructions. Cells were examined for GFP expression 24 and 48 hrs after transfection. Transfection experiments were performed a minimum of three times for each plasmid construct.
Media and cell lysates were collected for Western blot 48 hours post-transfection. Cells were lysed with RIPA buffer (150 mM NaCl, 50 mM Tris-HCl, 1% Triton-X100, 0.5% sodium deoxycholate, 0.1% SDS, 0.1 mM PMSF). Cell media was immunoprecipitated for Ab using 6 E10 antibody (Signet) and Gamma-Bind Plus Sepharose beads (Amersham/GE Healthcare). Western blot was used to assay GFP, C100 and Ab levels in cell homogenates and immunoprecipitated media. Briefly, between 25-200 mg of cell homogenate and immunoprecipitated media was added to SDS loading buffer (166 mM Tris-HCl, 8% sodium dodecylsulfate, 4% glycine, 2.5% 2-mercaptoethanol, pinch of phenol red, pH 6.8), boiled at 95uC for 10 min and loaded onto tris-trycine polyacrylamide gels. Proteins were separated using electrophoresis and transferred to nitrocellulose membranes at 250 mA overnight at 4uC. Membranes were blocked with 5% skim milk in TBS for 1 hr at room temperature and immunoblotted for GFP (1:3,000; Sigma) and C100/Ab (WO2; 1:2,500; kindly provided by Professor Colin Masters at University of Melbourne, Vic, Australia). Membranes were also blotted for bactin (1:20,000; Abcam) to ensure consistent protein loading. Protein bands were detected using enhanced chemiluminescence (ECL; Amersham Biosciences) and developed onto film. 50 mg of brain protein homogenate from APP SWE transgenic mouse brain was used as a positive control for Ab detection.

Surgery and Tissue Processing
Eight week old male C57Bl6/J mice were anesthetised by an intraperitoneal (i.p.) injection (10 ml/g) of a 1:1 mixture of ketamine (100 mg/ml) and xylazine (20 mg/ml) that was diluted 1:10 in sterile saline. Mice were secured in a stereotaxic frame and a small section of the skull was removed using a dental drill at the co-ordinates of 1.9 mm rostral and 1.3 mm lateral relative to lambda. For cerebellar injections a burr hole was drilled at approximately 1 mm caudal to lambda and 2 mm lateral to the midline. 1 ml of rAAV2 vector (10 12 GC/ml) or vehicle (phosphate buffered saline; PBS, pH 7.4) was directly injected into the hippocampus (1.5 mm ventral from the dura) and cerebellum using a pulled glass capillary at a rate of 0.2 ml/min. After each injection, the glass capillary was left in place for an additional 2 minutes before withdrawal. The skull fragment was then replaced and the skin closed using 4.0 sutures (Ethilon; Johnson & Johnson, Australia). After each experimental procedure mice received a subcutaneous injection of buprenorphine (0.02 mg/kg; Temgesic; Reckitt & Colman, Hull, UK) and an intramuscular injection of Benacillin (25 ml, Troy Laboratories Pty. Ltd. Australia).
The number of animals included in each group and the type of molecular analysis carried out at 3 months post-injection is shown in Table 1. Analysis of cerebellar rAAV2-C100-GFP injections was not possible due to problems with injection and thus consistent transduction of neurons in the cerebellar cortex. rAAV2-GFP, rAAV2-C100 V717F -GFP and rAAV2-Ab42-GFP were used as representative groups for further analysis at 6 months postinjection.
For cryosectioning, mice received an overdose of sodium pentobarbitone (i.p.; Lethabarb, Virbac) and were transcardially perfused with PBS (pH 7.4) containing 0.1% heparin followed by 4% paraformaldehyde in 0.1 M phosphate buffer (pH 7.4). Brains were post-fixed in 4% paraformaldehyde for 2 hours, cryoprotected by sinking in 30% sucrose and incubated in increasing concentrations of Jung freezing medium, all at 4uC. Brains were frozen in 100% Jung Freezing medium by immersion into isopentane on a bed of dry ice and stored at -80uC until sectioning. Serial coronal cryosections were cut at a thickness of 25 mm for slides 1-13 of each series (for immunohistochemistry) and a thickness of 14 mm for slides 14-15 of each series (for PCR), resulting in each slide containing between 5-6 serial sections, which were approximately 350 mm apart. Every section through the hippocampus and cerebellum was collected.
For Western blot and ELISA, mice received an overdose of sodium pentobarbitone (i.p.). The injected hippocampus, contralateral hippocampus, cerebellum and rest of the brain were collected individually, snap frozen in liquid nitrogen and stored at 280uC until homogenisation. Brain samples were homogenised in a volume of ice-cold PBS containing protease inhibitor (Roche) three times the weight of the tissue sample. Total protein was quantified using the Micro BCA protein assay kit (Pierce) and samples were stored at 280uC.
For immunohistochemistry, cryosections were washed with PBS, blocked for 1 hr in blocking buffer (PBS; 10% normal goat serum; 0.2% Triton-X100) and incubated overnight (4uC) with primary antibodies diluted in blocking buffer. Sections were washed with PBS, incubated with secondary antibodies diluted in blocking buffer for 2 hrs at room temperature, counterstained with Hoechst 33342 (Sigma; 4 ml/ml in PBS) and coverslipped with Dako fluorescent mounting media. Images were collected using a multiphoton laser scanning confocal microscope.
Immunoreactivity for GFAP, IBA-1 and MAP2 was semiquantified as staining density in the injected hippocampus and cerebellum. Low magnification confocal images were collected of the injected and contralateral hippocampi and cerebellar hemispheres. Three serial sections per brain in each brain region were analysed, the central section containing the injection site to maintain consistency between animals. Images were collected using the lowest magnification that allowed for adequate visualisation of immuno-positive staining. Thus, GFAP images were collected at 56magnification, IBA-1 images were collected at 106magnification and MAP2 images were collected at 206magnification. Because of the higher magnification necessary to image MAP2 staining, images of the hippocampus were collected from both the dentate gyrus and CA3 to ensure that the entire transduced region was imaged. Every image for each immunohistochemistry stain was collected using the same imaging conditions. GFAP, IBA-1 and MAP2 staining density was quantified using ImageJ analysis software (NIH). The area for quantification of hippocampal staining was defined by the hippocampal anatomical boundaries and density was measured in the injected and contralateral hippocampi. In the cerebellum the analysis region was defined by a rectangular area containing the GFP positive transduced region in the injected hemisphere and the corresponding region of the same area in the contralateral hemisphere. The staining density of the injected hippocampus and cerebellar hemisphere was calculated as a percentage of the corresponding contralateral hippocampus or cerebellar hemisphere and averaged across three sections per brain region. While it is acknowledged that injection of rAAV2 vectors could cause distal changes in contralateral brain regions, which would not be accounted for using this quantification method, the staining density was calculated this way in an attempt to standardise staining intensity across multiple immunohistochemistry runs. Furthermore, general observation of immunoreactivity in the contralateral hemisphere showed minimal distal effects and no individual differences between rAAV2 vectors and vehicle control, suggesting that contralateral differences would not bias results.

Ab Immunohistochemistry
Multiple antibodies against Ab were trialled including; antimouse 6E10 (Signet), anti-rabbit pan Ab (Zymed), anti-mouse 4G8 (Signet) and biotinylated Ab40 and Ab42 antibodies (gifts from Dr. Pankaj Mehta, NYS Institute for Basic Research, New York, USA). Extensive optimisation of the immunohistochemistry protocol was also performed, trialling heat and formic acid antigen retrieval methods either alone or in combination, multiple blocking reagents including 10% goat serum, 20% fetal bovine serum, 0.2% bovine serum albumin and mouse-on-mouse blocking reagent (M.O.M; Vector Laboratories) and both fluorescent and peroxidase immunohistochemistry. Positive control tissue sections from 18 month old APP SWE transgenic mice were used in every Ab immunohistochemistry trial.

IgG Staining and Quantification
Mouse IgG was stained by incubation with anti-mouse secondary antibodies. Briefly, sections were washed with PBS, blocked for 1 hr in blocking buffer (PBS; 10% normal goat serum; 0.2% Triton-X100) and incubated with anti-mouse secondary antibodies diluted in blocking buffer for 2 hrs at room temperature (anti-mouse Cy3; 1:300; Jackson or anti-mouse FITC; 1:100; MP Biochem). Sections were counterstained with Hoechst 33342 (Sigma; 4 ml/ml in PBS) and coverslipped with Dako fluorescent mounting media. Mouse IgG positive cells were identified by their distinctive morphology of intense circular IgG staining closely surrounding a nucleus. All mouse IgG positive cells were counted in three serial sections per brain in the hippocampus and the cerebellum, with the injection site as the central section for each brain region. For semi-quantitative analysis of IgG staining, we used sections immunostained for NeuN followed by a Cy3 anti-mouse secondary antibody. This allowed for consistent intensity grading of NeuN immunoreactivity versus non-specific IgG staining by comparing label in the injected hemisphere with the contralateral hemisphere. IgG staining intensity was graded in all serial sections through the hippocampus and cerebellum on a scale of 0-3, based on the relative intensity of extracellular IgG staining to positive staining in the surrounding densely packed neuronal layers (including the dentate gyrus, the pyramidal cell layer of CA regions and the granule cell layer in the cerebellum). Grading definitions: 0; no excess IgG staining, 1; stronger IgG staining intensity in neuronal layers than in the surrounding extracellular space, 2; same IgG staining intensity in the neuronal layers and extracellular space, 3; stronger IgG staining intensity in the extracellular space than in neuronal layers.

Western Blot
Total levels of C100, Ab and GFP protein were measured using the Western blot described above. 200 mg of total protein homogenate from each brain region was loaded and proteins were detected using the following primary antibodies; GFP (1:3,000; Sigma), C100/Ab (6E10; 1:1,000; Signet) and b-actin (1:20,000; Abcam).

Ab ELISA
Ab40 and Ab42 levels were measured in the injected hippocampus and cerebellum after injection of rAAV2-GFP, rAAV2-C100 V717F -GFP or rAAV2-Ab42-GFP. Protein was extracted from tissue samples using a previously published method designed to maximise the amount of Ab detection using ELISA [16]. Ab40 and Ab42 levels were quantified using the INNO-BIA plasma Ab forms test (Innogenetics) according to manufacturer's instructions. This kit combines ELISA and multiplex technology to assay Ab40 and Ab42 levels in the same sample, allowing high sensitivity detection of Ab using a small volume of sample. All samples were assayed in triplicate. A small amount of non-specific background protein detection was observed in the Ab42 assay results. Therefore, we conservatively deemed that any values that were greater than two standard deviations outside this background level to contain significant levels of Ab42.

PCR
RNA from the injected hippocampus, contralateral hippocampus and transduced cerebellar hemisphere was extracted from cryosections adjacent to those used for immunohistochemistry. Tissue was collected from five serial sections in the hippocampus and three serial sections in the cerebellum, with the injection site as the central section in each brain region. Injected and contralateral hippocampi and the transduced cerebellar hemisphere were carefully removed from whole brain sections using a scalpel blade and were collected individually. RNA was extracted from tissue samples using the FFPE RNeasy miniprep kit (Qiagen) according to the manufacturer's instructions and nucleic acid quality was assessed using a Nanodrop spectrophotometer. Between 300 ng and 400 ng of RNA was used in reverse transcription reactions performed using the Quantitect RT kit, which included a genomic DNA removal step (Qiagen).
Each qPCR reaction was performed using IQ PCR mix (BioRad) with 2 ml cDNA template, 5 mM of each primer and 1 ml of sterile water to a final volume of 10 ml, using a RotorGene RG 6000 PCR machine. Primer sequences and specific qPCR conditions are shown in Table 2. All runs were repeated, and values for each sample averaged. All PCRs were initially validated by agarose gel electrophoresis to check product specificity and size, with correct single bands excised from gels and purified using a Gel Purification kit (Qiagen) and sequenced using Big Dye Terminator v3.1 mix (Applied Biosystems) to confirm product identity. L19 and PPIA housekeeping genes were used for qPCR analysis and it was ensured that normalisation to each of these housekeeping genes resulted in similar expression patterns. Statistics All statistics were performed using SPSS (Version 17). Twotailed independent t-tests were initially used to determine if there was a significant difference between PBS and AAV2-GFP control groups in each brain region for all measures. The Levene's test for equality of variances was also performed to ensure normality of the distribution and to test the equality of variances assumption. As there was no difference observed between brain regions injected with PBS and rAAV2-GFP for any measure, rAAV2-GFP was used as the only control group for all further analysis, as this was the most appropriate control for comparison with experimental rAAV2 vectors. Univariate, one-way ANOVA was used when comparing more than two groups and the Bonferroni post-hoc test was used to determine individual significant differences between groups. The non-parametric Kruskal-Wallis Test was used to perform statistics on the intensity grading results from IgG staining. In all cases p,0.05 was deemed significant.

In vitro Production of C100 and Ab
To confirm that rAAV2 transgenes were translated into protein, HEK 293T, PC12 and HeLa cells were transfected with C100-GFP, C100 V717F -GFP, Ab40-GFP and Ab42-GFP plasmids. Strong GFP protein expression was observed using microscopy and Western blot ( Figure 1A), indicating that transfection with all plasmids was successful in each cell type. Transfection with C100-GFP and C100 V717F -GFP plasmids resulted in abundant intracellular C100 protein expression ( Figure 1A). Immunohistochemistry using the 6E10 antibody confirmed that C100 protein was only present in GFP-positive transfected cells, with immunopositive staining observed throughout the cytoplasm in a punctate pattern (data not shown).
No intracellular Ab protein could be detected using Western blot ( Figure 1A) or immunohistochemistry after transfection with all plasmids, in any cell type. Immunoprecipitation of Ab from the media of transfected cells showed that Ab was secreted from cells transfected with C100-GFP and C100 V717F -GFP plasmids, but we did not detect any secreted Ab from cells transfected with Ab40-GFP and Ab42-GFP plasmids ( Figure 1B).

In vivo Transduction of Hippocampus and Cerebellum
rAAV2-C100-GFP, rAAV2-C100 V717F -GFP, rAAV2-Ab40-GFP and rAAV2-Ab42-GFP vectors were injected into the hippocampus and cerebellum of 8 week old wild-type male mice. Transduction and resulting pathology was examined at 3 and 6 months post-injection. Injection of all AAV2 vectors resulted in wide-spread transduction in both brain regions, as shown by strong GFP expression (Figure 2A-C). Transduction efficiency appeared similar at 3 and 6 months post-injection.
Hippocampal injection resulted in transduction of all major hippocampal sub-regions including the dentate gyrus ( Figure 2G-I), CA regions of Amon's horn ( Figure 2D-F) and the subiculum. There was also evidence of anterograde and retrograde transport of rAAV2 and/or GFP as a result of hippocampal injection, as shown by GFP expression in brain regions distant from the injection site including the entorhinal cortex, contralateral hippocampus ( Figure 2J-L) and the anterior cingulate area of the cerebral cortex. Hippocampal rAAV2 injection resulted in transduction across a distance of up to 2.5 mm in the rostro-caudal plane.
Injection into the cerebellar cortex predominantly resulted in transduction of Purkinje cells ( Figure 2M-O), while transduction was also observed in the granule cell layer to a lesser extent. There was also evidence of anterograde and retrograde transport of rAAV2 and/or GFP with GFP expression observed in the deep cerebellar nuclei, the granule cell layer of the contralateral cerebellar cortex and in the external cuneate nucleus in the medulla. Cerebellar AAV2 injection resulted in transduction across a distance of up to 1.75 mm in the rostral-caudal plane.
All AAV2 vectors specifically transduced neurons in both the hippocampus and cerebellum, which was determined by morphology and by co-localisation with neuronal markers NeuN (for examples see Figure 2F, I, L, O) and MAP2 (data not shown). Furthermore, there was no co-localisation of GFP with GFAP or IBA-1, markers of astrocytes and microglia respectively, indicating that glia were not transduced by rAAV2 vectors (data not shown), a finding supported by previous studies [17].
To ensure that transduced brain regions also expressed the transgenes for the desired APP fragments in addition to GFP, qPCR was used to quantify C100-GFP, C100 V717F -GFP, Ab40-GFP and Ab42-GFP transgene mRNA levels in the injected Figure 1. C100 and Ab protein production in vitro after plasmid transfection. Analysis of C100 and Ab protein production in vitro after transfection with plasmids expressing Ab40-GFP, Ab42-GFP, C100-GFP or C100 V717F -GFP in cell homogenates (A) and media (B). C100 protein was only detected in cell homogenates after transfection with C100-GFP and C100 V717F -GFP plasmids. No Ab was detected in cell homogenates after transfection with any plasmid. APP SWE brain homogenate was used as a positive control for Ab detection. Strong GFP protein expression indicated successful transfection using all plasmids and b-actin was used as a loading control. Immunoprecipitation of Ab from the cell culture media revealed Ab in the media from cells transfected with C100-GFP and C100 V717F -GFP plasmids. Control refers to non-transfected cells. doi:10.1371/journal.pone.0059166.g001 Figure 2. Transduction in the hippocampus and cerebellum as indicated by GFP expression. Low magnification images show widespread GFP protein throughout the hippocampus (A) and cerebellar cortex (C). GFP protein was also observed in the contralateral hippocampus (B). Double hippocampus, contralateral hippocampus and the injected cerebellar hemisphere. RNA was extracted from fixed cryosections adjacent to those used for immunohistochemistry and GFP localisation. Strong mRNA expression of the human APP fragments was observed in transduced regions of the hippocampus and cerebellum at 3 and 6 months post-injection in all mice injected with C100 and Ab transgenes ( Figure 3A). There was also lower, but detectable expression of C100 and Ab mRNA in the contralateral hippocampus, providing evidence for retrograde transport of rAAV2 vectors and not just GFP protein. This finding supports previous studies that have shown neurons can transport AAV particles by retrograde axonal transport and that AAV then has the capability to express transgenes in distal brain regions [18,19]. There was qualitatively similar transgene mRNA expression in the hippocampus and cerebellum, indicating a similar level of transduction in both brain regions. Importantly, no human Ab or C100 mRNA expression was observed in any brain region of mice injected with rAAV2-GFP or PBS.
Human C100 and Ab protein expression was also examined using Western blot and immunohistochemistry. C100 immunopositive staining was observed throughout the cytoplasm and in primary processes of transduced neurons in the hippocampus and cerebellum ( Figure 3B-G). C100 protein was also detected using Western blot in hippocampal and cerebellar tissue injected with rAAV2-C100 V717F -GFP ( Figure 3H). In contrast, Ab could not be detected in either transduced brain region using immunohistochemistry trialling multiple antibodies against Ab or Western blot after injection with any rAAV2 vector (data not shown). These results led to the hypothesis that Ab may be present at very low levels and hence undetectable using Western blot or immunohistochemistry. Therefore, the high sensitivity INNO-BIA plasma Ab forms test was used to measure Ab40 and Ab42 levels in the injected hippocampus and cerebellum. Ab42 was detected in cerebellar tissue from every animal injected with rAAV2-C100 V717F -GFP, with Ab42 levels on average 5.964.3 (6 standard deviation) fold higher than baseline levels. Ab42 protein was also present in the cerebellum of one of the four animals injected with rAAV2-Ab42-GFP, this animal having Ab42 levels that were 3.3 fold higher than baseline levels. Ab42 protein was also detected in the hippocampus from animals injected with either rAAV2-C100 V717F -GFP or rAAV2-Ab42-GFP but Ab42 protein levels did not reach the two standard deviation criterion relative to baseline. It is important to note that, because the same viral vectors were injected into each brain region, the lack of definitive detection of Ab42 in the hippocampus suggests that Ab42 may be processed differently in each brain region.
Significant levels of Ab40 were not found in the cerebellum or hippocampus after injection of either rAAV2-C100 V717F -GFP or rAAV2-Ab42-GFP, although a relatively high level of Ab40 protein expression was seen in hippocampal tissue from one rAAV2-C100 V717F -GFP injected animal. The inter-animal variation in the amount of Ab detected, and the differences in Ab levels between the cerebellum and hippocampus, leads to speculation that Ab was produced but was then degraded or cleared after production, resulting in overall low levels available for detection in the murine brain samples.

Microgliosis
Immuno-positive staining density of the microglial marker IBA-1 was analysed to determine the extent of microgliosis following injection of rAAV2 vectors. This analysis technique was designed to examine wide-spread microgliosis across the whole transduced region. IBA-1 positive microglia in areas of gliosis showed morphological signs of activation including enlarged cell bodies, ramified processes and increased IBA-1 staining within individual microglia ( Figure 4G, K).
In the cerebellum, injection of rAAV2-Ab42-GFP also resulted in a significantly greater level of microgliosis at 3 months postinjection than after injection of rAAV2-GFP (Bonferroni; p,0.05; Figure 4N). Microgliosis was no longer elevated at 6 months postinjection. Therefore, combined results from the hippocampus and cerebellum suggest that microgliosis was strongly associated with Ab42 rather than Ab40 expression, as it was most extensive after injection with rAAV2-Ab42-GFP and rAAV2-C100 V717F -GFP.

Astrogliosis and Neuronal Density
Ab expression has been associated with astrogliosis and neuronal death, therefore the extent of astrogliosis and neuronal density were examined using immunohistochemistry for GFAP and MAP2 respectively. Quantification of GFAP staining density showed that while there was some evidence of increased astrocyte activation surrounding the injection site in both brain regions following injection of rAAV2-C100-GFP, rAAV2-C100 V717F -GFP, rAAV2-Ab40-GFP and rAAV2-Ab42-GFP in comparison to after injection of rAAV2-GFP or PBS, this increase was not extensive enough to be significantly different at either 3 or 6 months post-injection (data not shown). Quantification of MAP2 staining density showed that neuronal density was not altered in the hippocampus or cerebellum as a result of injection of any rAAV2 vector at either 3 or 6 months post-injection (data not shown).

Permeability of the Blood Brain Barrier
An unexpected consequence of injection with rAAV2 vectors was the presence of increased mouse IgG staining around the injection site, first detected after immunohistochemistry using antimouse secondary antibodies. Further testing showed that the increased IgG staining was observed when sections were incubated only with anti-mouse secondary antibodies, was localised to transduced brain regions, and was not observed after injection with PBS, therefore showing that it was not a technical artefact of the immunohistochemistry procedure.
Increased IgG staining was observed in the hippocampus and cerebellum after injection of rAAV2 vectors expressing APP fragments, but not after injection of rAAV2-GFP or PBS ( Figure 5). The relative extracellular IgG staining intensity was graded in the hippocampus and cerebellum after immunohistochemistry for NeuN, using the IgG staining intensity of the surrounding densely packed neuronal layers as the staining intensity comparison. IgG staining intensity in the hippocampus at 3 months post-injection was significantly higher after injection of rAAV2-C100-GFP (Kruskal-Wallis; p = 0.009), rAAV2-C100 V717F -GFP (Kruskal-Wallis; p = 0.001) and rAAV2-Ab42-GFP (Kruskal-Wallis; p = 0.016) than after injection of AAV2-GFP. There was also a non-significant trend for increased IgG staining in the hippocampus after injection of rAAV2-Ab40-GFP than after injection of AAV2-GFP at 3 months post-injection (Kruskal-Wallis; p = 0.056). Similarly, IgG staining intensity in the cerebellum was significantly higher at 3 months post-injection with rAAV2-C100 V717F -GFP (Kruskal-Wallis; p = 0.014), rAAV2-Ab40-GFP (Kruskal-Wallis; p = 0.014) and rAAV2-Ab42-GFP (Kruskal-Wallis; p = 0.013) than after injection with rAAV2-GFP. In contrast, IgG staining intensity was no longer significantly elevated at 6 months post-injection with any rAAV2 vector in either the hippocampus or cerebellum, indicating that the increased IgG in the brain decreased with time.
Staining for mouse IgG also strongly labelled small, round cells that were predominantly observed in transduced regions of the hippocampus and cerebellum (Figure 6-A-C). Immunohistochemistry revealed that these IgG positive cells were also immunopositive for IBA-1 ( Figure 6D-G), a calcium binding peptide produced specifically by monocytes and activated microglia [20]. The number of IgG positive cells was counted and averaged across three serial sections in the hippocampus and cerebellum. The vast majority of IgG positive cells were observed throughout the injected hippocampus and cerebellum and in the lateral ventricles adjacent to the injected hippocampus, with a small number of cells also observed in the contralateral hemispheres in both brain regions. Almost no IgG positive cells were seen after control PBS injections. 3 months post-injection, there were significantly more IgG positive cells in the hippocampus after injection with rAAV2-C100-GFP, rAAV2-C100 V717F -GFP and rAAV2-Ab42-GFP than after injection with rAAV2-GFP ( Figure 6H; Bonferroni; p,0.001). In comparison, while the number of IgG positive cells in hippocampi at 3 months post-injection with rAAV2-Ab40-GFP was higher than after rAAV2-GFP injection, this difference was not statistically significant. There was no longer a significant difference between groups in the number of IgG positive cells in the hippocampus at 6 months post-injection ( Figure 6H).
The number of IgG positive cells was also increased in the cerebellum after injection with rAAV2 vectors containing APP fragments, but high inter-animal variation meant that these increases were not statistically significant at 3 months postinjection (Bonferroni; Figure 6I). However, analysis using the less stringent LSD post-hoc test found that injection with rAAV2-C100 V717F -GFP resulted in significantly more infiltrating cells than injection with rAAV2-GFP (LSD; p = 0.049). This effect was still observed at 6 months post-injection (Bonferroni; p,0.05).

Gene Expression
PS1, apoE, clusterin and synaptophysin gene expression was examined in the injected hippocampus, contralateral hippocampus and injected cerebellar hemisphere at 3 and 6 months post-injection. ApoE and clusterin expression was also examined to determine if expression of genes involved in Ab clearance was altered [21]. Synaptophysin is a synaptic protein marker and was therefore examined as an additional marker of neurodegeneration and PS1 expression was examined to determine if C100 mediated endogenous PS1 expression, hence altering production of Ab. Altered levels of each of these genes and the resulting proteins have also been previously associated with AD [22,23,24,25]. Differences in gene expression between groups were only deemed significant if the statistical difference was consistently observed after normalisation to two housekeeping genes; L19 and PPIA. It was found that there was not a significant difference in expression of any gene examined as a result of injection of any vector at 3 or 6 months post-injection in either the hippocampus or cerebellum.

Discussion
Injection of rAAV2 vectors expressing transgenes for human Ab40, Ab42, C100 and C100 V717F 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-Ab42-GFP and rAAV2-C100 V717F -GFP in the hippocampus resulted in significantly increased IBA-1 density at 3 months post-injection, while injection of rAAV2-Ab40-GFP and rAAV2-C100-GFP did not, suggesting that Ab42 may be a more potent mediator of microgliosis than Ab40. It is known that Ab attracts and causes activation of microglia [26,27], but it has been previously suggested that this may only occur in response to fibrillar Ab [1,28,29,30]. The association between fibrillar Ab 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 [31,32], and manipulating the amount of plaque deposition results in similar changes in the extent of microgliosis [33,34,35]. However, the results from this study suggest that microgliosis can also occur in response to soluble forms of Ab, 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 postinjection. 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 [36]. 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 [37,38,39,40,41]. 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 [42]. 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 [43], in agreement with previous studies [37,44].
In the hippocampus at 3 months post-injection, blood brain barrier disruption was more extensive after exposure to Ab42, via . Ab and C100 mRNA and protein expression in the hippocampus and cerebellum. mRNA for Ab and C100 transgenes was detected using PCR (A). Immunostaining for C100 using CT20 antibody revealed C100 protein in GFP positive neurons (B-G). C100 protein was also detected using western blot (H). HIP (INJ): injected hippocampus, HIP (CONTRA) contralateral hippocampus, CB: cerebellum, ROB: rest of the brain. Lanes 1, 4, 7, 10; representative brain injected with AAV2-C100 V717F -GFP, lanes 2, 5, 8, 11; representative brain injected with AAV2-Ab42-GFP, lanes 3, 6, 9, 12; representative brain injected with AAV2-GFP. doi:10.1371/journal.pone.0059166.g003 expression of Ab42 directly or the C100 and C100 V717F precursors. In comparison, while IgG staining intensity and numbers of infiltrating cells after injection with rAAV2-Ab40-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 [45,46] and previous studies have hypothesised that it may be directly caused by Ab [42,47,48]. The results from this study not only support the hypothesis that Ab expression may directly cause blood brain barrier disruption, but also suggest that Ab42 may be a more potent mediator of blood brain barrier disruption than Ab40.
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 Ab 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 Ab to cause activation and migration of astrocytes [49,50,51]. However, it has also been shown that the activation of astrocytes in AD is dependent upon the conformation and aggregation state of Ab; astrocytes surrounding dense core plaques become activated, while astrocytes surrounding diffuse plaques or those not associated with aggregated Ab do not show signs of activation and can often show signs of atrophy [52,53]. Therefore, it is possible that the lack of extensive astrogliosis observed in this study was due to the lack of aggregated, fibrillar Ab following injection of viral vectors. The lack of widespread neurodegeneration observed in this study is consistent with previous studies that have shown that Ab is not a potent mediator of neurodegeneration in vivo [54]. 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 Ab [55,56,57]. It is interesting that synaptophysin mRNA levels were unaffected as Ab, particularly soluble oligomers of Ab, has been shown to decrease expression of synaptic markers, including synaptophysin [32,58,59,60,61]. However, this finding is not consistent across all studies [62,63], indicating that decreased synaptophysin expression may not be a direct consequence of Ab 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 Ab40 and Ab42 protein expression in transduced brain regions. We do not believe that this was due to the use of inefficient viral vectors because Ab42 was detected in tissues injected with rAAV2-C100 V717F -GFP or rAAV2-Ab42-GFP using the more sensitive ELISA technique, thus proving that these vectors were capable of producing Ab. Furthermore, there was strong transduction of all bi-cistronic 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 Ab protein translation. Finally, injection of rAAV2-Ab40-GFP and rAAV2-Ab42-GFP resulted in the development of obvious brain pathology that was unique to each Ab isoform and was not present after injection of vehicle or rAAV2-GFP controls, strongly suggesting that the Ab produced as a result of rAAV2 vector injection was responsible for the observed pathologies. We hypothesise that the inability to detect Ab using the less sensitive immunohistochemistry and western blotting methods, and the observation of variance between animals in the amount of Ab detected using ELISA, was due to rapid clearance and/or degradation of Ab. Ab 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 Ab levels. Indeed, some of the pathology observed in vivo including microgliosis and infiltration of cells from the periphery indicate that Ab degradation may have been increased in response to expression of Ab as both microglia and infiltrating monocytes are capable of Ab phagocytosis and enhance degradation [44,64,65,66,67,68]. Increased Ab clearance may also explain the lack of Ab40 detected using the ELISA-mutiplex assay, as this isoform is more readily cleared and degraded than Ab42 [69,70]. Furthermore, the Indiana mutation promotes the preferential production of Ab42 rather than Ab40, therefore the presence of this mutation and the hypothesised increased clearance of Ab could account for the very low levels of Ab40 observed after injection of rAAV2-C100 V717F -GFP. Lack of available tissue prevented ELISA-based quantification of Ab40 levels after injection of rAAV2-Ab40-GFP.
An intriguing finding was that, while significant levels of Ab42 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 Ab42 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 Ab. 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 Ab [71,72], and that plaques do eventually appear in the cerebellum as AD pathology advances [73], indicating that the cerebellum is capable of producing amyloid pathology. Nonetheless, the cerebellum consistently has fewer plaques and lower levels of insoluble Ab and intracellular Ab42 [73,74,75] 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 [76]. 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 Ab [77,78]. The present data suggest an alternative hypothesis, that cerebellar cells may be intrinsically less responsive to the presence of Ab 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 Ab than vectors directly expressing Ab transgenes, both in vitro and in vivo. This most likely resulted from the more physiological method of production of Ab from C100, in comparison to the non-physiological production by direct expression of either Ab40 or Ab42. Direct expression of Ab may not be optimal for Ab accumulation, possibly due to Ab production occurring in the incorrect sub-cellular location. Previous in vitro studies have shown that fusing Ab and C100 to a signal protein that directs expression in the secretory pathway greatly increases the amount of Ab detected after plasmid transfection [79,80], hence suggesting that sub-cellular location of Ab 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 Ab and as a result became better equipped to deal with the consequent pathology, such as by increasing levels of Ab degrading enzymes or increasing antiinflammatory proteins. However, further studies are necessary to confirm this hypothesis.
In conclusion, the use of viral vectors to over-express Ab and C100 is a promising technique with which to examine the consequences of Ab expression in mature CNS tissues in vivo. Results from this study provide evidence that Ab42 causes greater pathology than Ab40, particularly by promoting microgliosis and inducing abnormal permeability changes in the blood brain barrier.