Astrocyte-derived clusterin suppresses amyloid formation in vivo

Accumulation of amyloid-β (Aβ) peptide in the brain is a pathological hallmark of Alzheimer’s disease (AD). The clusterin (CLU) gene confers a risk for AD and CLU is highly upregulated in AD patients, with the common non-coding, protective CLU variants associated with increased expression. Although there is strong evidence implicating CLU in amyloid metabolism, the exact mechanism underlying the CLU involvement in AD is not fully understood or whether physiologic alterations of CLU levels in the brain would be protective. We used a gene delivery approach to overexpress CLU in astrocytes, the major source of CLU expression in the brain. We found that CLU overexpression resulted in a significant reduction of total and fibrillar amyloid in both cortex and hippocampus in the APP/PS1 mouse model of AD amyloidosis. CLU overexpression also ameliorated amyloid-associated neurotoxicity and gliosis. To complement these overexpression studies, we also analyzed the effects of haploinsufficiency of Clu using heterozygous (Clu+/−) mice and control littermates in the APP/PS1 model. CLU reduction led to a substantial increase in the amyloid plaque load in both cortex and hippocampus in APP/PS1; Clu+/− mice compared to wild-type (APP/PS1; Clu+/+) littermate controls, with a concomitant increase in neuritic dystrophy and gliosis. Thus, both physiologic ~ 30% overexpression or ~ 50% reduction in CLU have substantial impacts on amyloid load and associated pathologies. Our results demonstrate that CLU plays a major role in Aβ accumulation in the brain and suggest that efforts aimed at CLU upregulation via pharmacological or gene delivery approaches offer a promising therapeutic strategy to regulate amyloid pathology.


Background
Alzheimer's disease (AD) is the most common form of age-related dementia, currently affecting more than 5 million individuals nationwide [1] . Given the large scale of AD and steady increase in the aging population, providing a better understanding of the pathogenesis of this disease is imperative. Deposition of amyloid-β (Aβ) peptide in the brain is a key initiating event leading to the development of AD [2]. Toxic amyloid aggregates accumulate in the brain in the form of extracellular plaques [3] and are commonly found in leptomeningeal and cortical blood vessels as cerebral amyloid angiopathy (CAA) [4,5].
Clusterin (CLU), also known as apolipoprotein J (apoJ), is a ubiquitous glycoprotein widely expressed throughout the human body, with very high expression levels observed in the nervous system [6]. As a prominent extracellular chaperone, CLU is involved in heterogenous biological processes, including lipid homeostasis, complement inhibition, cell cycle, and apoptosis [7]. Large-scale genome-wide association studies (GWAS) have identified a significant association of common polymorphisms within the CLU gene with risk of developing AD [8,9]. Initial studies have reported the ability of CLU to form complexes with Aβ [10] and influence its solubility [11][12][13], thus preventing amyloid fibril formation. In vitro reports have further shown the protective role of CLU against amyloid-mediated neurotoxicity [14]. Subsequently, we and others have demonstrated that complete CLU deletion using global knockout (Clu −/− ) mice in different models of amyloidosis has a profound effect on amyloid aggregation and clearance, resulting in reduced total and fibrillar plaques [15][16][17][18][19]. However, as with all studies utilizing completenull alleles, unknown developmental or compensatory factors could cloud our interpretation of these data with Clu −/− mice.
Recent studies have suggested that CLU upregulation may play a protective function. Increased CLU levels have been reported in AD-vulnerable brain regions [20] and in the cerebrospinal fluid (CSF) of AD patients [21]. In addition, the association of elevated levels of plasma CLU with severity of AD has also been shown [21]. Importantly, the protective T allele of the major CLU variant (rs11136000) has been associated with increased CLU levels [8,22], suggesting a beneficial effect of CLU upregulation. However, other studies have reported conflicting data showing a significant correlation of increased CLU with brain atrophy and rapid clinical progression of AD patients [23]. Thus, it remains unclear whether elevated CLU levels represent a neuroprotective function in AD.
In the present study, we used different in vivo approaches to show that both~30% CLU overexpression in the brain or a~50% reduction have prominent effects on amyloid pathology and gliosis. Specifically, sustained increase of CLU expression in astrocytes via viral delivery in the APP/PS1 mouse model of amyloidosis led to amelioration of amyloid accumulation and Aβ-mediated neurotoxicity and gliosis. We further demonstrated that CLU haploinsufficiency exacerbates amyloid deposition and gliosis in APP/PS1 mice. These findings may have important implications for optimization of amyloidrelated treatments in AD, especially strategies aimed at altering the levels of CLU protein.

Methods
Animals APP/PS1 mice aged 8 months and bearing a double mutation in APP and PS1 (APPswe/PS1ΔE9) were used in this study [24]. Littermate breeding strategy of Clu +/− mice bred to APP/PS1; Clu +/− mice was also used. Mice were housed in a temperature and humidity-controlled environment under a 12-h light/dark cycle and with free access to food and water. All studies were performed in accordance with National Institute of Health Guide for the Care and Use of Laboratory Animals (National Research Council (2011) Guide for the Care and Use of Laboratory Animals (National Academies Press, Washington, DC), 8th Ed.) under the approved protocol from the Mayo Clinic Institutional Animal Case and Use Committee. All analyses included mice of both sexes in accordance with National Institute of Health directives.

AAV-GFP and AAV-CLU viral production
Viral vector construction and AAV production was performed, as previously described [25]. Briefly, CLU or GFP expression plasmids were cloned into an AAV vector. The constructs were sequence-verified using ABI3730 with Big Dye chemistry (Applied Biosystems, Foster City, CA). AAV vectors expressing GFP and CLU under the control of the glial fibrillary acidic protein (GFAP) promoter to drive expression in astrocytes were co-transfected with AAV2/8 helper plasmids into HEK293T cells. Cells were harvested and lysed in the presence of 0.5% sodium deoxycholate and 50 U/ml Benzonase (Sigma, St. Louis, MO) by freeze-thawing 48 h post-transfection, and the virus was isolated using a discontinuous iodixanol gradient. Quantitative PCR was used to measure the genomic titer of each virus.

Intracerebroventricular injections
AAV-GFP or AAV-CLU viruses were injected bilaterally into cerebral lateral ventricles of APP/PS1 and WT pups at postnatal day 2 with 2.75E+ 10 viral particles/ventricle. Briefly, postnatal day 2 mice were cryoanesthesized on ice for 5 min via a cold metal plate. The skull was pierced with the 30-gauge needle just posterior to Bregma and 2 μl of AAV virus was injected into the lateral ventricles. Following the injections, pups were placed on the warm pad until they regained normal color and resumed movement.

Histological analysis
To examine CLU association with amyloid in humans, a brain specimen was obtained from Mayo Clinic Brain Bank from an individual with pathologically confirmed AD (Braak stage 6, Thal stage 5). Paraffin-embedded sections were first cut at 10 μm, then deparaffinized, rehydrated, and subjected to antigen-retrieval in dH 2 O for 15 min under high temperature. Brain sections were incubated overnight with goat anti-CLU antibody (1:50, Santa Cruz) diluted in 0.5% dry-milk in PBS, washed three times in PBS-X and PBS, followed by overnight incubation with secondary antibody (1:250, Jackson Immu-noResearch). Sections were double labeled with the thioflavine-S stain to detect fibrillar amyloid and mounted using Vectashield (Vector Laboratories, Inc.). The images were captured using Zeiss LSM 700 laser scanning confocal microscope.
For histopathological analyses of amyloid accumulation in mouse brain, 8-month-old APP/PS1 AAV-GFP and APP/PS1 AAV-CLU mice, and APP/PS1; Clu +/+ and APP/ PS1; Clu +/− mice were used. Mice were deeply anesthetized with pentobarbital (100 mg/kg i.p.) and transcardially perfused with phosphate buffered saline (PBS) to expunge blood from cerebrovasculature. After brain removal, cortex and hippocampus from one hemibrain were isolated, frozen on dry ice, and stored at − 80°C until further processing.

Presparation of brain lysates
Brain lysates were prepared using a 3-step sequential extraction, as previously described [26]. Briefly, cortex and hippocampus were homogenized in 500 μL and 300 μL, respectively, of Tris-buffered saline (TBS) containing protease and phosphatase inhibitor cocktail. The brain regions were briefly sonicated followed by the ultracentrifugation at 100,000×g for 1 h at 4°C. The supernatant was transferred to a new Eppendorf tube and stored in − 80°C as TBS fraction (soluble fraction). The cortical and hippocampal pellets were resuspended in 500 μL and 300 μL, respectively, of TBS with 1% Triton X-100 (TBS-X) with protease and phosphatase inhibitor cocktail, briefly sonicated, and incubated for 30 min at 4°C with gentle rotation. The incubation was followed by ultracentrifugation at 100,000×g for 1 h at 4°C. The supernatant was transferred to a new Eppendorf tube and stored in − 80°C as TBS-X fraction. The final cortical and hippocampal pellets were resuspended in 300 μL and 100 μL, respectively, of 70% formic acid (FA), followed by rotation for 2 h at 4°C and ultracentrifugation at 100,000×g for 1 h at 4°C. The final supernatant was neutralized by the addition of 20 volumes of 1 M Tris base and stored in − 80°C as FA fraction (insoluble fraction).

Quantitative analysis of CLU and Aβ levels by ELISA
CLU protein levels in the cortex and hippocampus were assessed by specific enzyme-linked immunosorbent assay (ELISA). Plates were coated using the capture antibody (Mouse Clusterin DuoSet ELISA, R&D Systems) and CLU was detected by using the detection antibody (Mouse Clusterin DuoSet ELISA, R&D Systems), followed by the incubation with Super Slow ELISA TMB reagent (Sigma). Total protein concentrations used for normalization were assessed by Bicinchoninic Acid (BCA) Protein Assay kit (Thermo Scientific), according to the manufacturer's instructions with a standard curve using BSA.
The levels of soluble and insoluble Aβ 40 and Aβ 42 were assessed by sensitive sandwich Aβ 40 or Aβ 42 -specific ELISAs. Plates were coated using human anti-Aβx-40 (13.1.1) and Aβx-42 (2.1.3) capture antibodies. Aβ standards were prepared by using human synthetic Aβ 40 and Aβ 42 . To detect Aβ species, HRP-conjugated Ab5 secondary antibody was used followed by the incubation with Super Slow ELISA TMB reagent (Sigma) to develop. Protein concentrations were assessed by Bicinchoninic Acid (BCA) Protein Assay kit (Thermo Scientific), according to the manufacturer's instructions with a standard curve using BSA.

Quantitative analysis of amyloid deposition and neuritic dystrophy
For stereological analyses 8-month-old APP/PS1 AAV-GFP and APP/PS1 AAV-CLU mice, and APP/PS1; Clu +/+ and APP/PS1; Clu +/− were used. As previously described [17], for each animal, 3 coronal sections, separated by 300 μm, were used for the quantification. The cortical and hippocampal regions were marked and StereoInvestigator software (MBF Bioscience) was used to count the percentage of area covered by total and fibrillar amyloid, CAA presence in leptomeningeal and penetrating vessels, and neuritic dystrophy in the cortex and hippocampus.

Real-time quantitative PCR
Total RNA was isolated using RNeasy Mini Kit (Qiagen) according to the manufacturer's instructions. Randomprimed reverse transcription was performed using High-Capacity cDNA Reverse Transcription kit

AAV-mediated CLU expression in astrocytes
To investigate the effect of differential CLU levels on amyloid pathology, we first evaluated the CLU localization in APP/PS1 mice and human AD brain. Consistent with previous studies [17], we observed an abundant immunoreactivity of CLU surrounding amyloid plaques in APP/PS1 mice (Fig. 1a) and human AD brain sections (Fig. 1a). We next delivered intracerebroventricular injections of adeno-associated viral (AAV) vectors expressing murine CLU (AAV-CLU) or green fluorescent protein (GFP, AAV-GFP) as control into APP/PS1 and wild-type (WT) littermates at postnatal day 2. Given that astrocytes are the major source of CLU in the brain, we specifically targeted CLU expression in astrocytes by using the glial fibrillary acidic protein (GFAP) promoter. Three months post viral injection we assessed the pattern of GFP expression in the brain. Immunofluorescent analysis showed a widespread GFP immunoreactivity throughout the brains of APP/PS1 (APP/PS1 AAV-GFP ) animals with high levels of GFP expression observed in cortical and hippocampal regions (Fig. S1). In addition, we found an extensive colocalization of GFP with GFAP-positive astrocytes ( Fig.  1b) but no co-expression of GFP within microglia (Fig.  1b) or neurons (Fig. 1b), confirming the specificity of viral transduction as has been reported previously with this paradigm [27]. We next evaluated CLU protein levels to determine the degree of overexpression we achieved in cortex and hippocampus of 8-month-old WT and APP/PS1 animals, measured by a CLU-specific enzyme-linked immunosorbent assay (ELISA). Viral injection of AAV-CLU led to a sustained but physiologic CLU overexpression in cortex of WT (*p < 0.05) and APP/PS1 (**p < 0.01) mice of~30% compared to animals injected with control AAV-GFP vectors (Fig. 1c). Similarly, a significant~35-40% increase in CLU levels was detected in hippocampus of WT AAV-CLU (**p < 0.01) and APP/PS1 AAV-CLU (**p < 0.01) mice relative to animals injected with control AAV-GFP virus (Fig. 1c). These data show that viral delivery of AAV-CLU leads to a specific yet physiologic overexpression of CLU in astrocytes that persists throughout life.

Increased CLU levels influence neurotoxicity and gliosis
Neuritic dystrophy, in the form of severely swollen dendrites and axons, is commonly observed around fibrillar plaques [28]. To determine whether increasing CLU levels influenced the formation of neuritic dystrophy, we performed histological examination with lysosomalassociated membrane protein 1 (Lamp1) labeling to mark dystrophic neurites in proximity to X-34 labeled fibrillar amyloid aggregates. Although Lamp1 immunoreactivity was observed around amyloid plaques in APP/ PS1 AAV-GFP and APP/PS1 AAV-CLU animals, CLU overexpression significantly reduced the overall amount of dystrophic neurites in cortex and hippocampus of APP/ PS1 AAV-CLU mice relative to their controls (Fig. 2e, f and  Fig. S4a). No obvious differences were seen in the number of dystrophic neurites around individual plaques between either group of mice (Fig. S4b).
Since abundant gliosis is associated with the presence of amyloid pathology [17], we next assessed whether CLU overexpression had a differential effect on inflammatory changes in APP/PS1 mice. Although reactive astrocytes and microglia, labeled with GFAP and IBA1 immunostaining, respectively, were present in close proximity to amyloid plaques in APP/PS1 AAV-GFP and APP/PS1 AAV-CLU mice (Fig. 3a, d), we found significant differences in the level of gliosis associated with CLU overexpression. Specifically, GFAP immunoreactivity was reduced in cortex of APP/PS1 AAV-CLU mice compared with APP/PS1 AAV-GFP mice (Fig. 3a, b and Fig.  S4c). Similarly, we found a markedly reduced expression of IBA1 in mice injected with AAV-CLU compared to controls (Fig. 3d, e Fig. S4e). However, the levels of gliosis were not different between groups when normalized to fibrillar amyloid (Fig. S4d, f). Finally, we tested whether CLU upregulation in astrocytes affects gliosis at the molecular level. We observed a significant reduction in the expression of Gfap and Cst7 (encoding cystatin F) transcripts further confirming the CLU effect on astrogliosis and microgliosis, respectively (Fig. 3c, f). Collectively, these results indicate that CLU overexpression is associated with reduction of both neuritic dystrophy and gliosis, which is accompanied by decrease in Aβ accumulation in APP/PS1 AAV-CLU mice compared with APP/ PS1 AAV-GFP mice.
Analysis of Lamp1 immunoreactivity revealed more dystrophic neurites surrounding amyloid plaques in brain parenchyma in APP/PS1; Clu +/− mice compared to APP/PS1; Clu +/+ animals (Fig. 4e, f and Fig. S7a). No differences between groups were observed in the levels of Lamp1 when normalized to fibrillar amyloid (Fig. S7b). Additionally, APP/PS1; Clu +/− showed a marked increase in astrogliosis assessed by the GFAP immunoreactivity compared with APP/PS1; Clu +/+ littermate controls (Fig. 5a, b and Fig. S7c). We did not observe a significant difference in the number of GFAP-positive astrocytes between APP/PS1; Clu +/+ and APP/PS1; Clu +/− mice after GFAP immunostaining was normalized to amyloid load (Fig. S7d). Similarly, the evaluation of IBA1 immunoreactivity showed significant differences between CLU genotypes, with CLU haploinsufficient mice having higher levels of microgliosis (Fig. 5d, e and  Fig. S7e). No significant differences in the levels of microgliosis between APP/PS1; Clu +/+ and APP/PS1; Clu +/− animals were observed when normalized to fibrillar amyloid (Fig. S7f). Finally, real-time quantitative PCR revealed an increase in the Gfap transcript level in APP/ PS1; Clu +/− animals compared to APP/PS1; Clu +/+ mice (Fig. 5c). Taken together, these data show that partial yet physiological reduction of CLU expression increases severity of amyloid pathology and exacerbates amyloidassociated neurotoxicity and gliosis.

Discussion
For over three decades the relationship between CLU and Aβ has been studied. Early discoveries of CLU upregulation in AD patients [20], its ability to interact with Aβ [10], and strong genetic association of the CLU variants with AD risk [8,9], have fueled subsequent studies focusing on the mechanism underlying the CLU contribution to AD pathogenesis. Despite the tremendous progress in our understanding of CLU biology, the complexity of its function in the context of AD is still not fully understood. This study was aimed to further elucidate the role of CLU in AD with the focus on physiologic manipulation of expression and the resulting impact on the pathological accumulation of amyloid in the brain.
This study is the first to evaluate the effect of CLU haploinsufficiency on pathological changes in the brain. We have shown that partial loss of CLU by~50% resulted in exaggerated amyloid accumulation in brain parenchyma with significant effects on neurotoxicity and gliosis. This agrees with previous studies showing that CLU increases Aβ solubility and prevents Aβ aggregation. Moreover, it has recently been suggested that the molar ratio of CLU to Aβ is a critical factor in determining Aβ aggregation, with a higher CLU:Aβ ratio promoting Aβ solubilization and lower CLU:Aβ ratio facilitating Aβ accumulation in the brain [34]. Interestingly, recent sequencing studies have found numerous nonsynonymous variants and small insertions/deletions within the CLU β-chain region (e.g. p.R338W, p.I360N, p.I303NfsX13, p.T445_D447del) [35]. In vitro studies, that aimed to elucidate the nature of these rare mutations enriched in AD patients, have reported associations between CLU variants and reduced levels of secreted CLU [36].
Taken together, our findings highlight a prominent function of CLU in regulating amyloid aggregation in the brain. In particular, we uncovered neuroprotective properties of CLU overexpression, that results in a significant reduction of amyloid pathology and overall improvement of amyloid-related pathological features, including neurotoxicity and gliosis. Moreover, our study is the first to show that CLU haploinsufficiency leads to more severe amyloidosis. Future investigations should further address how CLU overexpression provides neuroprotection in Aβ-mediated disorders.

Conclusions
No effective treatments to prevent or slow the progression of AD have been developed to date. However, Aβ has long been considered a promising target for therapeutic interventions. Our study provides a strong evidence that~30% CLU overexpression ameliorates amyloid pathology while~50% reduction of CLU exacerbates amyloid accumulation in the brain, thus reflecting a protective role of CLU in AD. Given that increasing CLU levels has potential consequences for Aβ-related therapies, future studies determining the exact mechanism and factors modulating CLU expression are critical. Our findings also indicate that pharmacological or gene delivery approaches aimed at increasing the levels of CLU in the brain could be a viable therapeutic strategy for combating AD. specimens. L. D, K.J.W., L.P. produced AAV viral vectors. G.B. provided reagents and help with data interpretation. All authors read and approved the final manuscript.