WNT5A Signaling Contributes to Aβ-Induced Neuroinflammation and Neurotoxicity

Neurodegenration is a pathological hallmark of Alzheimer's disease (AD), but the underlying molecular mechanism remains elusive. Here, we present evidence that reveals a crucial role of Wnt5a signaling in this process. We showed that Wnt5a and its receptor Frizzled-5 (Fz5) were up-regulated in the AD mouse brain, and that beta-amyloid peptide (Aβ), a major constituent of amyloid plaques, stimulated Wnt5a and Fz5 expression in primary cortical cultures; these observations indicate that Wnt5a signaling could be aberrantly activated during AD pathogenesis. In support of such a possibility, we observed that inhibition of Wnt5a signaling attenuated while activation of Wnt5a signaling enhanced Aβ-evoked neurotoxicity, suggesting a role of Wnt5a signaling in AD-related neurodegeneration. Furthermore, we also demonstrated that Aβ-induced neurotoxicity depends on inflammatory processes, and that activation of Wnt5a signaling elicited the expression of proinflammatory cytokines IL-1β and TNF-α whereas inhibition of Wnt5a signaling attenuated the Aβ-induced expression of the cytokines in cortical cultures. Our findings collectively suggest that aberrantly up-regulated Wnt5a signaling is a crucial pathological step that contributes to AD-related neurodegeneration by regulating neuroinflammation.


Introduction
Beta-Amyloid (Ab) peptide is a dominant candidate of the causative agents for Alzheimer's disease (AD) [1,2]. According to the widely-held amyloid hypothesis of AD, Ab initiates an array of molecular and cellular cascades that eventually lead to progressive neuronal dysfunction and degeneration [1,2]. However, mechanistic molecular processes that link Ab and neurodegeneration remain to be firmly established.
Chronic neuroinflammation associated with persistent glial activation is a major disease process evoked by Ab and intimately associated with the progress of AD pathologies [3,4]. Previous studies suggest that neuroinflammation contributes to the development of neurodegenerative hallmarks in AD brains, including Ab plaques [4] and tau tangles [5,6,7]. AD therapeutic approaches that target neuroinflammation are under development [8,9,10,11]. AD neuroinflammation is likely triggered by Ab-mediated activation of microglia and astrocytes [3,4,12,13,14]. It was reported that Ab induces the expression of cytokines (including IL-1b, TNF-a, IL-6, and IL-8) in cultured astrocytes and microglia [15,16,17,18]. Mounting evidence suggests that Ab may activate glial cells via specific sensor receptors such as toll-like receptors (TLR), receptors for advanced glycoxidation end-products (RAGE) and NOD-like receptors (NLR) [4]. Despite the significant understandings on the induction of AD neuroinflammation, the downstream molecular processes that are elicited by Ab and regulate the inflammation remain poorly understood.
Wnts are secreted signaling proteins that play important roles in neural development and plasticity [19,20,21,22]. Multiple lines of evidence indicate a critical role of Wnt signaling in AD [22]. bcatenin, a key downstream effector protein in the canonical Wnt signaling pathway, interacts with and is regulated by presenilin [23,24,25]. Glycogen synthase kinase (GSK)-3, a central serine/ threnine kinase in the canonical Wnt signaling pathway, plays a critical role in the regulation of Ab production [26] and aggregation [27] and in tau phosphorylation [28]. Genetic studies revealed that LRP6 polymorphisms are causally linked to AD [29]. In AD brains, canonical Wnt signaling is impaired [30], and DKK1, an antagonist of Wnt signaling, is upregulated [31,32]. Importantly, Ab was reported to inhibit Wnt signaling by directly binding to the Frizzled receptors [33]. The impairment of canonical Wnt signaling is likely etiologically significant, because forced up-regulation of the canonical Wnt signaling pathway has rescuing effects on the development of AD-related phenotypes in both neuron cultures and animal models [27,30,34,35]. In contrast to the canonical pathway, the involvement of non-canonical Wnt signaling pathways in the regulation of AD pathogenesis is less clear. A recent study indicates that Wnt5a-activated non-canonical Wnt signaling antagonizes Ab synaptotoxicity [36].
In this paper, we report an important role of Wnt5a signaling in the regulation of Ab-evoked neurotoxicity and neuroinflammation. We observed that (1) Wnt5a/CaMKII signaling is upregulated at the early stages of AD development in an APPswe/ PSEN1DE9 transgenic mouse model, (2) Ab activates Wnt5a signaling in primary cortical cultures, (3) Ab-induced Wnt5a upregulation is a critical molecular step leading to the development of Ab neurotoxicity in cultures, (4) Wnt5a stimulates inflammatory processes, and (5) Wnt5a is critical for Ab-induced inflammatory response. Our results suggest that abnormally up-regulated noncanonical Wnt5a signaling may regulate chronic neuroinflammation in AD brains.

Preparation of Ab42 peptides
Recombinant human Ab (1-42) (Chemicon) was used to prepare monomers (Ab-mon), oligomers (Ab-olig) and fibrils (Ab-fib), as described [37]. For monomer preparation, 5 mM Ab42 in DMSO was diluted directly with cell culture media. For oligomer preparation, 5 mM Ab42 in DMSO was diluted to 100 mM in ice-cold D-MEM/F-12 (Invitrogen), vortexed for 300, centrifugated (10,000 g; room temperature) for 19, and incubated at 4uC for 24 h. For fibril preparation, 10 mM HCl was added to the Ab42 solution (5 mM in DMSO) to bring Ab to a final concentration of 100 mM, vortexed for 300, centrifugated (10,000 g, room temperature) for 19, and incubated with shaking at 37uC for 24 h. Products from such preparations are mixtures that contain Ab monomers and the intended aggregates [37]. The quality of the preparations was confirmed by Western blotting analysis with the 6E10 (Covance; 1:5000), and the molecular mass was estimated by rainbow pre-stained protein markers (GE Healthcare).

Animals and hippocampus dissection
All procedures were approved by Animal Care and Use Committees of the University of Texas Medical Branch. Male APP (amyloid precursor protein)/presenilin-1 double transgenic (APPswe/PSEN1DE9, 26Tg) mice [38] and wild-type littermates at 3.5 months of age were anesthetized and then decapitated. The hippocampi were rapidly collected on ice for Western blotting, immunohistochemistry, or fluorescent immunostaining..

Neuron cultures
C57BL E18 cortical cultures were prepared, as described [39]. The cells were plated on poly-D-lysine (30,000-70,000; Sigma) -coated dishes at a density of 1.5610 5 cells/cm 2 . Immunostaining with cell-type markers indicate that the cultures are mixtures of neuron (,60%) and glial cells (,40%). Cultures at 12-14 DIV were used in this study. All experimental treatments were carried out by adding the administrated agents into the freshly changed culture media. Only morphologically healthy neuronal cultures were used for drug treatments.

Statistical analysis
Data were expressed as the means 6 SEM. Statistical analysis was performed with the Prism software (GraphPad). We used the Student's two-tailed t-test for statistical comparisons between any two groups, and one-way ANOVA analyses with a Bonferroni post-hoc test for comparisons between multiple groups.

Increased expression of key proteins in the non-canonical Wnt signaling pathway in the hippocampus of AD mouse models
Previous studies revealed down-regulation of canonical Wnt signaling in the brain of APPswe/PSEN1DE9 AD mouse models [27]. Consistent with these findings, we observed decreased expression of proteins in the canonical pathway, including Wnt3a, Fz1 and b-catenin (data not shown). In contrast, in preliminary studies, we also observed up-regulated expression of several proteins implicated in the non-canonical Wnt signaling pathway, including Wnt5a and its receptor Fz5, at various postnatal stage (3.5-9.5 months). At 3.5 months of age, APP transgene was clearly expressed (Fig. 1A). APP processing expected to generate Ab was evident in the transgenic hippocampi (Fig. 1B). In addition, immunohistochemistry staining with an antibody that recognized Ab revealed clustered signals in the cell body layer (Fig. 1C), which may include Ab that starts accumulating. Because memory deficits have not manifested at this stage [44], we reasoned that the upregulation of non-canonical Wnt signaling is potentially an early molecular abnormality during AD pathogenesis. Therefore, we performed more detailed analysis on the non-canonical Wnt signaling in mice at 3.5 months of age. Western blotting analysis showed that, compared with wild-type controls, Wnt5a protein increased 1.5 fold in the hippocampus of the 26Tg mouse (p,0.01; Fig. 2A). Similarly, Fz5 protein was also significantly upregulated in the 26Tg hippocampus (1.6 fold; p,0.01; Fig. 2B). Because activation of non-canonical Wnt signaling can cause aCaMKII phosphorylation [45], we next determined the protein level of phosphorylated aCaMKII (pT286-aCaMKII). pT286-aCaMKII was increased 2.6 fold (p,0.05; Fig. 2C). Fluorescent immunostaining showed that, in comparison with control, the increase of Wnt5a (Fig. 2D) and Fz5 (Fig. 2E) occurred in all hippocampal fields, including CA1, CA2/CA3 and dentate gyrus (DG), of the 26Tg mouse, which resembled the spatial pattern of Ab plaques (Fig. 1C). These data together suggest that noncanonical Wnt signaling is up-regulated in the 26Tg AD hippocampus.
Rapid up-regulation of non-canonical Wnt signaling by Ab oligomers in primary neuron culture We next sought to understand the mechanism by which noncanonical Wnt signaling is activated in the AD mouse brain. Because Ab is a key pathological agent for AD, we hypothesized that it causes the up-regulated expression of Wnt5a and other relevant proteins measured in 26Tg mice. To test this hypothesis, we determined the effect of Ab on the expression of these proteins in mouse cortical cultures (12)(13)(14). A wide range of Ab concentrations (100 nM -15 mM) have been used in previous in vitro studies. We chose to use a relatively low Ab concentration (500 nM) to better reflect the fact that the proteins are upregulated at early AD stages when Ab begins to accumulate (Fig. 1C). Prior works have shown that 500 nM Ab is efficient to initiate some of the earliest and reversible AD-relevant pathologies [46,47]. More recent studies have indicated that different forms of Ab (monomer, oligomer and fibril) vary in their AD-related toxicity, with the oligomer being considered the most toxic form [37,48]. We prepared the oligomers and fibrils of Ab (1-42) (Fig. 3A) and compared the effect of Ab monomers, oligomers and fibrils on Wnt5a, Fz5 and pT286-aCaMKII protein expression in cortical cultures. As shown in Fig. 3B-D, Ab oligomer treatment for 24 h caused significant increases in Wnt5a (1.4 fold; p,0.05), Fz5 (1.4 fold; p,0.01) and pT286-aCaMKII (1.7 fold; p,0.01). Under the same experimental conditions, preparations of monomers and fibrils caused changes with lower magnitudes (Fig. 3B-D). Next, we investigated the dose effects of the oligomer. As shown in Fig. 3E-G, the oligomer increased expression of Wnt5a, Fz5, and pT286-aCaMKII proteins in a concentrationdependent manner (50 nM-5 mM). Although Wnt5a and p-CaMKII displayed a clear dose-effect at 24 hours after treatment, Fz5 showed this effect at 30 min. In addition, 500 nM of Ab oligomer caused a time-dependent increase in Wnt5a (Fig. 3H).  These results demonstrated that the non-canonical Wnt5a signaling is up-regulated by Ab peptide, especially the oligomers.

Suppression of Wnt5a signaling alleviates Ab neurotoxicity
The observations of the upregulation of the Wnt5a signaling in early AD mouse brains and by Ab in cortical cultures suggest a potential role of Wnt5a signaling in AD pathogenesis. We next sought to directly test this idea by determining the significance of Ab-mediated Wnt5a up-regulation on Ab cytotoxicity in cortical cultures. Previous work revealed concentration-dependent neurotoxicity of Ab peptides [37,49,50], with high concentrations of Ab (.5.0 mM) causing neuron death in cultures and low concentrations of Ab (0.1,1.0 mM) sensitizing neurons to become vulnerable to further stresses [50]. To mimic the early AD brain condition when the non-canonical Wnt signaling proteins are aberrantly upregulated, in these experiments we used a relatively low concentration (500 nM) of Ab that can upregulate Wnt5a and Fz5 (Fig. 3). Ab oligomer at this concentration did not cause cell death, as measured by MTT assays (Fig. 4A). We also used trypan blue staining to confirm this observation (Fig. S1). This finding is consistent with the idea that in the early stages of AD pathogenesis Ab may not cause neuron death. Next, we wanted to investigate if Ab at this concentration causes the cells to become more vulnerable to detrimental stresses that are relevant to AD pathogenesis. To this end, we tested the effect of Ab on the sensitivity of the cultures to hydrogen peroxide (H 2 O 2 ) stresses. Hydrogen peroxide was chosen because it is generated during the very early stages of aggregation of the amyloid peptides [51] and is critically involved in AD pathogenesis [49,52]. At low concentrations (25-100 mM), hydrogen peroxide did not cause cell death by itself; however, when co-administrated with Ab oligomers (500 nM) it caused marked cell death (Fig. 4A, Fig. S1). These results indicate that neurons exposed to 500 nM Ab are more vulnerable to other detrimental factors. We also showed that H 2 O 2 neither stimulated aCaMKII phosphorylation by itself nor significantly enhanced the Ab-induced phsphorylation (Fig. 4F), indicating that H 2 O 2 did not activate the non-canonical pathway.
To determine the potential role of Wnt5a up-regulation in Abinduced sensitization to H 2 O 2 stress, we neutralized Wnt5a in media with specific anti-Wnt5a antibody. As shown in Fig. 4C, the antibody (0.5-8.0 mg/ml) displayed a rescue effect on cell death in a concentration-dependent manner, which reached statistical significance after 4.0 mg/ml. Recent studies identified a modified Wnt5a-derived hexapeptide (Box5) that can specifically antagonize Wnt5a in malignant melanoma cells [53]. Similar to the anti-Wnt5a antibody, Box5 also had a rescue effect on Ab/H 2 O 2induced cell death in a concentration-dependent manner (Fig. 4E). Because Fz-5 is a Wnt5a receptor [54], we further tested the effect of anti-Fz5 antibody (against the extracellular cystein-rich domain). Consistent with the findings with anti-Wnt5a antibody and Box5, the anti-Fz5 antibody also had a rescue effect (Fig. 4C). We further tested the effect of, anti-Wnt5a, anti-Fz5 antibodies and Box5 on the phosphorylation of aCaMKII, a molecular marker of the activation of the non-canonical Wnt/Ca 2+ pathway. The results showed that all these reagents significantly blocked Ab/H 2 O 2 -induced up-regulation of p-aCaMKII (Fig. 4G), and thus provide an independent molecular confirmation of the activity of anti-Wnt5a, anti-Fz5 antibodies and Box5. These results together strongly suggest that Ab-induced Wnt5a upregulation at least partially mediates Ab neurotoxicity.

Exogenous Wnt5a potentiates Ab neurotoxicity
Because Ab-induced Wnt5a up-regulation is critical for Ab neurotoxicity (Fig. 4C and E), we next investigated if exogenous Wnt5a can potentiate Ab/H 2 O 2 -induced cell injury. To this end, we tested the effect of recombinant Wnt5a on Ab/H 2 O 2 neurotoxicity. As shown in Fig. 4B, compared with Ab/H 2 O 2treated cultures, addition of purified recombinant Wnt5a protein (50-800 ng/ml) increased cell death in a concentration-dependent manner. This potentiation activity of Wnt5a was completely abolished when the protein was heat-inactivated (Fig. 4B). This observation indicates that the biochemical activity of Wnt5a rather than the mere presence of Wnt5a peptide is essential for the potentiation. In addition, Foxy5, a Wnt5a-derived hexapeptide (the same peptide as Box5 but differently modified) that can mimic Wnt5a-induced activities in malignant melanoma and in breast cancer cells [55], was also able to potentiate the neurotoxicity of Ab/H 2 O 2 (Fig. 4D). These results indicate that Wnt5a is sufficient to elicit neurotoxicity.

Anti-inflammation factors attenuate Ab neurotoxicity
The results described above indicate that up-regulated Wnt5a may mediate Ab neurotoxicity. We then sought to understand the mechanism by which Wnt5a contributes to Ab neurotoxicity. Because recent studies revealed a critical role of Wnt5a in the regulation of inflammatory responses in peripheral systems, we were interested in determining if Wnt5a contributes to Ab neurotoxicity by regulating inflammation. To this end, we first investigated the contribution of neuroinflammation to the Ab-induced sensitization of cortical cultures to H 2 O 2 stress. We found that IL-10, a prototypic antiinflammatory cytokine, caused concentration-dependent rescue effects on Ab toxicity (Fig. 5). IL-10 started to rescue Ab/H 2 O 2induced cell death at 0.25 ng/ml and reached a plateau effect at 1 ng/ml (Fig. 5A). Heat-inactivated IL-10 did not have any rescuing effect (Fig. 5A). To further confirm the involvement of inflammation in the Ab toxicity, we determined the effect of activated protein C (APC), another well established antiinflammatory protein. The results showed that APC displayed a similar rescue effect as IL-10 (Fig. 5B). These results indicate that neuroinflammatory response likely contributes to Ab neurotoxicity.

Wnt5a regulates proteins that involve in inflammatory response in cortical cultures
Recent studies reveal an important role of Wnt5a in control of inflammatory response in non-neuronal peripheral tissues [56,57,58], but its involvement in regulation of inflammatory response in nervous systems has not been reported. To investigate the potential contribution Wnt5a signaling to neuroinflammation, we examined the effect of Wnt5a on the protein expression of NF-kB-inducing kinase (NIK), a key positive regulator of inflammation that controls NF-kB activity by promoting the processing of p100, the inactive NF-kB2 precursor, to produce the functional p52 subunit [59]. We observed that NIK protein began to increase 5 min after Wnt5a administration (200 ng/ml), and this increase continued gradually, peaked at 60 min. and stayed at a higher level afterward (Fig. 6A). Heat-inactivated Wnt5a did not induce a NIK increase (Fig. 6B), indicating that Wnt5a protein activates inflammatory pathways in cortical cultures. We also investigated the effect of Wnt5a on the expression of IkB-a protein, an inhibitory regulator of inflammation response by sequestering NF-kB in the cytosol [60]. Interestingly, Wnt5a also induced a gradual increase of IkB-a (Fig. 6C). In contrast to the NIK increase, the IkB-a up-regulation did not begin until 30-60 min after Wnt5a treatment but continued for the rest of the time that measurements were recorded (Fig. 6C). It is possible that the lag in up-regulation of IkB-a functions to resolve the Wnt5a-initiated inflammatory response.
Inflammatory response culminates with the secretion of an array of inflammatory cytokines. If Wnt5a indeed induces the inflammatory response, specific inflammatory cytokines must be secreted after Wnt5a administration. Thus, we used ELISA to measure IL-1b and TNF-a in culture media after Wnt5a treatment for 0.5, 1, 2, 6, 12, or 24 hrs. We observed that Wnt5a stimulation caused a 2.7-fold increase in IL-1b at 12 h, while no change was detected at other time points (Fig. 7A). On the other hand, TNF-a increased significantly at 6 h (2.3 fold), 12 h (4.3 fold) and 24 h (3.1 fold) (Fig. 7C). However, this effect of Wnt5a was completely abolished when the protein was heat-inactivated ( Fig. 7B and D). These results together strongly suggest that Wnt5a can elicit the inflammatory response in neuron cultures.

Discussion
In this study, we found that non-canonical Wnt5a signaling is up-regulated in mouse brains prior to AD phenotypes and by Ab peptide in cortical neuron cultures. The up-regulated Wnt5a signaling contributes to the inflammation-dependent Ab neurotoxicity in cultures. We also found that Wnt5a up-regulates inflammation regulatory proteins and proinflammatory cytokines and that Wnt5a is required for the Ab-induced proinflammatory cytokines. These observations collectively suggest the following working model (Fig. 8C): accumulation of Ab in the brain aberrantly up-regulates Wnt5a signaling, which in turn evokes an inflammatory response that causes neurodegeneration or cell death in AD brains.
The observed up-regulation of Wnt5a signaling is probably an early etiologically relevant event during AD development. Both Wnt5a and Fz5 proteins significantly increase in the APPswe/ PSEN1DE9 hippocampus at the age of 3.5 months (Fig. 2). Previous studies showed that this AD mouse model started to accumulate Ab plaques after 4 months of age [61] and did not develop cognitive impairments until 5-7 months of age [44]. Thus,  the observed Wnt5a and Fz5 up-regulation at 3.5 months of age is likely prior to the development of major AD phenotypes. This notion is consistent with the finding that a relatively low concentration of Ab is able to up-regulate Wnt5a and Fz5 (Fig. 3), suggesting that Wnt5a signaling is a potential target for slowing or blocking early AD pathogenesis.
Converging lines of evidence support a critical role of the downregulation of the canonical Wnt/b-catenin pathway in AD pathogenesis. In contrast, the involvement of non-canonical Wnt signaling is less clear. Our findings reveal an early up-regulation of Wnt5a signaling in the hippocampus of 26Tg AD mice. Etiological significance of this dysregulation is suggested by the observation that Wnt5a signaling is necessary for Ab to fully induce neurotoxicity in cortical cultures. Previous studies demonstrated that down-regulation of canonical signaling contributed to Ab neurotoxicity [27,30,34,62]. It is possible that Ab causes parallel up-regulation of the non-canonical Wnt signaling and down-regulation of the canonical signaling to initiate neurotoxicity cascades. The Wnt canonical and non-canonical pathways often antagonize one another [63]. Thus, another possible scenario is that Ab may directly down-regulate the canonical pathway, as suggested by a recent study [33], which consequently causes the up-regulation of the non-canonical pathway.
Our results reveal a neurotoxic activity of Wnt5a signaling, and this Wnt5a activity contributes to Ab toxicity in neuron cultures. Cerpa et al. recently reported that acute administration of exogenous Wnt5a (500 nM) prevented Ab-induced synaptotoxicity within 40 min after Ab oligomer application [36]. Their results indicate a synapto-protective activity of Wnt5a signaling soon after Ab exposure. Because Ab-up-regulated Wnt5a does not occur by 1 hour after Ab treatment in cultures (Fig. 3H) and 500 nM Ab itself does not induce obvious cell death in this period ( Fig. 4A and S1), we reason that basal Wnt5a has a synaptoprotective activity. On the other hand, sustained up-regulation of Wnt5a, which occurs at 2 hours after Ab treatment (Fig. 3H), probably potentiates neurotoxicity.
We further found that activation of Wnt5a signaling stimulates the expression of proinflmmatory cytokine in cortical cultures ( Figs. 6 and 7). This finding indicates that up-regulation of Wnt5a may mediate Ab-induced neuroinflammation in AD brains (Fig. 8C). Because the Ab-elicited inflammatory response (Fig. 8A, 8B) and alleviated Ab-induced neurotoxicity (Fig. 4) was impaired by the anti-Wnt5a antibody and Box5, Ab likely induces Wnt5a secretion, although the kinetics of the secretion is currently unknown. In peripheral non-neuronal systems, Wnt5a is implicated in inflammation of multiple chronic disorders, including rheumatoid arthritis [64], sepsis [56], atherosclerosis [65], melanoma [66], and psoriasis [67]. Our results provide the initial evidence for a critical role of Wnt5a signaling in the regulation of inflammatory responses in CNS disorders. Because the primary cortical cultures used in this study contain neurons and glia (including microglia and astrocytes), we currently do not know the specific type of glial cells through which Wnt5a evokes the observed inflammatory responses. In a recent study, Halleskog et al. reported that Wnt3a stimulated the expression of proinflammatory cytokines in microglia [68]. It would be interesting to know if Wnt5a regulates neuroinflammation by stimulating the same or different types of glia. Nonetheless, the findings on Wnt5a and Wnt3a indicate that proteins in the Wnt family may orchestrate neuroinflammatory response during AD pathogenesis.

Supporting Information
Figure S1 Cell survival rates revealed by trypan blue staining. Cortical cultures at 24 hrs after indicated treatments were used. Dying cells were stained due to the increase of membrane permeability to the dye.