β2 Adrenergic Receptor, Protein Kinase A (PKA) and c-Jun N-terminal Kinase (JNK) Signaling Pathways Mediate Tau Pathology in Alzheimer Disease Models*

Background: Accumulating evidence indicates that β receptors (βAR) may be involved in Alzheimer disease (AD) pathology and that amyloid β peptide (Aβ) may interact with β2AR independently of presynaptic activities. Results: β2AR, PKA, and JNK mediate Aβ-induced phosphorylation of tau in vivo and in vitro. Conclusion: An Aβ-β2AR signaling is involved in tau pathology in AD. Significance: This work indicates a potential mechanism for altering AD pathology by blocking β2ARs. Alzheimer disease (AD) is characterized by neurodegeneration marked by loss of synapses and spines associated with hyperphosphorylation of tau protein. Accumulating amyloid β peptide (Aβ) in brain is linked to neurofibrillary tangles composed of hyperphosphorylated tau in AD. Here, we identify β2-adrenergic receptor (β2AR) that mediates Aβ-induced tau pathology. In the prefrontal cortex (PFC) of 1-year-old transgenic mice with human familial mutant genes of presenilin 1 and amyloid precursor protein (PS1/APP), the phosphorylation of tau at Ser-214 Ser-262 and Thr-181, and the protein kinases including JNK, GSK3α/β, and Ca2+/calmodulin-dependent protein kinase II is increased significantly. Deletion of the β2AR gene in PS1/APP mice greatly decreases the phosphorylation of these proteins. Further analysis reveals that in primary PFC neurons, Aβ signals through a β2AR-PKA-JNK pathway, which is responsible for most of the phosphorylation of tau at Ser-214 and Ser-262 and a significant portion of phosphorylation at Thr-181. Aβ also induces a β2AR-dependent arrestin-ERK1/2 activity that does not participate in phosphorylation of tau. However, inhibition of the activity of MEK, an upstream enzyme of ERK1/2, partially blocks Aβ-induced tau phosphorylation at Thr-181. The density of dendritic spines and synapses is decreased in the deep layer of the PFC of 1-year-old PS1/APP mice, and the mice exhibit impairment of learning and memory in a novel object recognition paradigm. Deletion of the β2AR gene ameliorates pathological effects in these senile PS1/APP mice. The study indicates that β2AR may represent a potential therapeutic target for preventing the development of AD.

Neurofibrillary tangles composed of hyperphosphorylated tau in the brain is a hallmark of Alzheimer disease (AD) 2 , and the phosphorylation of tau may be a major pathological cause of the disorder by inducing synapse loss (1)(2)(3)(4). Increasing evidence suggests that soluble amyloid ␤ peptide (A␤) is linked to hyperphosphorylation of tau at serine and threonine residues (5,6). A recent study has demonstrated that A␤ causes tau to wander into dendrites, leading to loss of synapses, spines, and microtubules (7)(8)(9). In 3xTg-AD mice harboring a knockin mutation for presenilin 1 (PS1, M146V) and transgenes for amyloid precursor protein (APPswe) and tau (tauP301L), spine loss occurs exclusively at dystrophic dendrites that accumulate both A␤ oligomers and hyperphosphorylated tau intracellularly (10), and it is the phosphorylation of tau that causes the protein to stray (11). Previous publications have shown that A␤ induces phosphorylation of tau at serine and threonine residues via a myriad of signaling cascades. However, little is known about how A␤ induces tau hyperphosphorylation and AD development.
In a recent epidemiological study, it was found that antihypertensive medication, including ␤ blockers, may reduce the risk of AD (12). Another survey in AD patients indicates that ␤ blockers may be associated with a delay of functional decline in the patients (13). There is also evidence that ␤ 2 AR may be involved in AD pathogenesis through effects on A␤ production and inflammation (14,15). Another study has shown recently that polymorphism in ␤ 2 AR contributes to sporadic late-onset AD, which may be related to the availability and response of ␤ 2 AR (13,16). Meanwhile, ␤ 2 AR also plays an important role in cognition and stress-related behaviors (17,18).
Recent studies have characterized that A␤ induces activation of ␤ 2 AR-mediated PKA-and G protein-coupled receptor * This work was supported by a new investigator research grant from the kinase/arrestin-dependent signal transduction, which is presynaptic activity-independent and requires the N terminus of ␤ 2 AR (19 -22). Although a prolonged treatment with A␤ induces GRK/arrestin 3-dependent internalization and degradation of ␤ 2 AR, which impairs presynaptic activity-dependent neurotransmission, the intracellular levels of cAMP and PKA activity are partially preserved, reaching a balance between receptor activation and degradation (23). Besides PKA, the internalization-associated arrestin signaling can trigger the phosphorylation of MAPK and JNK that may phosphorylate tau (24), and the activation of the exchange protein activated by cAMP (Epac) may also mediate JNK phosphorylation linked to tau (25). In this study, we aim to understand the significance of ␤ 2 AR signaling cascades in tau pathology in AD.
In some experiments, inhibitors for kinases and receptors were added as indicated 10 min before administration of A␤.
Golgi Staining-An FD Rapid GolgiStain TM kit (MTR Scientific, MD) was used to stain dendritic spines of neurons in the deep layer of the PFC of 1-year-old and 6-month-old wild-type, ␤ 2 -KO, PS1/APP, and ␤ 2 -KO/PS1/APP mice. Briefly, animals were perfused with heparinized PBS and 2% paraformaldehyde, followed by an additional perfusion with PBS to wash away excessive PFA in the body. Brains were dissected out and stained with Golgi-Cox impregnation solutions. After staining, the brains were sliced at a thickness of 240 m on a LEICA Vibratome 1000. The slices were dehydrated and mounted on slides. Images were taken using a Carl Zeiss LSM-700 microscope equipped with DIC objective lenses. All spines observable along 100-m dendritic segments at least 25 m from the cell soma were counted.
Immunofluorescence Microscopy-Wild-type, ␤ 2 -KO, PS1/ APP, and ␤ 2 -KO/PS1/APP were perfused consecutively in vivo with heparinized PBS and 2% PFA. The brains were dissected out and post-fixed with 2% PFA overnight. After serial dehydration in sucrose, the brains were frozen in Tissue-Tek O.C.T compound (VWR LabShop, IL), and slices were cut at a thickness of 40 m on a CM3050 S cryostat (Leica Microsystems, Inc., Germany). Brain slices and fixed primary neurons were blocked and permeabilized with goat serum and Nonidet P-40 in PBS and then incubated with primary antibodies. Alexa Fluor 488-or Alexa Fluor 568-conjugated secondary antibodies (Invitrogen) were used to reveal the primary antibodies. Nuclei were counterstained with DAPI (Thermo Scientific, IL). Quantification of synapsin I positively stained synapses was performed with the Analyze Particles commands of the Fiji software.
Novel Object Recognition Test-The task was carried out according to previous publications (27,28). The experimental apparatus consisted of a Plexiglas open-field box (40 ϫ 40 ϫ 29 cm). The apparatus was placed in a sound-isolated room. The novel object recognition task procedure consisted of three sessions: habituation, training, and retention sessions. Each mouse was habituated individually to the box with 10 min of exploration in the absence of objects. During the training session in the next day, two objects (A and B) were placed in the back corner of the box, 10 cm from the side wall. A mouse was then placed in the middle front of the box, and the total time spent in exploring the two objects was recorded for 10 min by the experimenter with two stopwatches. Exploration of an object was defined as directing the nose to an object at a distance of less than 2 cm and/or touching it with the nose. During the retention session on the third day (24 h after the training session), the animals were placed back into the same box, in which one of the familiar objects was replaced by a novel object, C. The animals were then allowed to explore freely for 10 min, and the time spent exploring each object was recorded. Throughout the experiments, the objects were used in a counterbalanced manner in terms of their physical complexity and emotional neutrality. A preference index, which is the ratio of the amount of time spent in exploration of any one of the two objects (training session) or the novel object (retention session) over the total time spent exploring both objects, was used to measure cognition.
Statistical Analyses-Unpaired Student's t test and one-or two-way analysis of variance was used to compare different groups with Prism software as indicated (GraphPad, CA). p Ͻ 0.05 was considered significant.

RESULTS
To explore the role of ␤ 2 AR signaling in tau pathology in relationship with A␤ in AD, we cross-bred the AD animal model overexpressing the human familial APPswe and PS1 mutants (PS1/APP) with mice lacking the ␤ 2 AR gene (␤ 2 -KO). We found that the phosphorylation of tau at Ser-214, Ser-262, and Thr-181 was increased in the PFC of 6-month-old and 1-year-old PS1/APP mice compared with wild-type mice (Fig.  1, A-C, and data not shown). However, deletion of the ␤ 2 AR gene abolished the increases in phosphorylation of tau at Ser-  ␤2AR/PKA/JNK Correlate A␤ and phorylation of tau at Thr-181 in the PFC of PS1/APP mice (Fig.  1, A-C, and data not shown). The phosphorylation of JNK1, GSK3␣/␤, and CaMK II was also increased in the PFC of 1-yearold PS1/APP animals, but the increases in phosphorylation of these proteins were greatly blunted in ␤ 2 -KO/PS1/APP mice (Fig. 1, D-F).
We then applied primary PFC neurons isolated from wildtype and ␤ 2 -KO animals to further dissect A␤-induced ␤ 2 AR signaling cascades in tau phosphorylation. A␤ (10 Ϫ6 M) induced tau phosphorylation at Ser-214, Ser-262, and Thr-181 in wildtype PFC neurons, but the increases in tau phosphorylation were almost abolished at Ser-214 and Ser-262 and blunted significantly at Thr-181 in ␤ 2 -KO neurons (Fig. 2, A-C). A minimal dose of 10 Ϫ8 M of A␤ was effective to promote tau phos-phorylation (data not shown). Meanwhile, a ␤ 2 AR-selective antagonist, ICI118551, blocked A␤-induced tau phosphorylation (Fig. 3, A-C). As a control, a general ␤AR agonist, isoproterenol, also induced robust increases in tau phosphorylation, which was inhibited by a ␤AR antagonist, alprenolol (Fig. 3, A-C). Moreover, the effects of A␤ on tau phosphorylation at Ser-214 and Ser-262 were blocked by inhibition of adenylyl cyclase with 2,5-dideoxyadenosine-3-triphosphate tetrasodium (10 Ϫ4 M, Fig. 3G) or inhibition of PKA inhibitor with membrane-permeable myristoylated PKI (10 Ϫ5 M, D and E). The A␤-induced tau phosphorylation was not affected by the Epac inhibitor brefeldin A (10 Ϫ7 M), and treating the cells with the Epac-selective activator 8-CPT-2Me-cAMP (10 Ϫ7 M) for 5 min did not induce tau phosphorylation (Fig. 3, H and I). These results indicate that tau phosphorylation at Ser-214 and Ser-262 is primarily dependent on ␤ 2 AR-adenylyl cyclase-PKA signaling. In comparison, A␤-induced tau phosphorylation at Thr-181 was only blocked partially by inhibition of PKA with PKI (10 Ϫ5 M, Fig. 3F).
In agreement with published literature, we found that the density of dendritic spines in the deep layer of the PFC in 1-year-old and 6-month-old PS1/APP mice was decreased. However, deletion of the ␤ 2 AR gene reversed the decrease (Fig.  8, A-C). Unlike the relatively even distribution of synapses in the deep layer of the PFC in WT mice, PS1/APP mice displayed regions with a dramatically decreased number of synapses, as indicated by synapsin I staining (Fig. 8, D and E), and surrounding synapses remained in clusters (arrows). Deletion of the ␤ 2 AR gene in PS1/APP mice yielded a distribution of synapsin I positively stained synapses similar to those in WT or ␤ 2 -KO mice (Fig. 8, D and E). To assess the cognitive role of ␤ 2 AR in PS1/APP transgenic AD animals, we tested learning and memory in 1-year-old mice in a novel object recognition paradigm. We found that PS1/APP mice showed impaired learning and memory, whereas ␤ 2 -KO/PS1/APP mice performed significantly better than PS1/APP mice (Fig. 9A). Knockout of the ␤ 2 AR gene itself tended to improve learning and memory in 1-year-old mice (Fig. 9A). However, it tended to impair learning and memory in 6-month-old mice (Fig. 9D). In the training  session, one-year-old animals in each group showed a similar preference for the reference object (Fig. 9B). The total exploration time in PS1/APP and ␤ 2 -KO/PS1/APP animals in the training session was similar, indicating similar locomotor activity in these mice (Fig. 9C).

DISCUSSION
Recent epidemiological studies suggest that ␤ blockers may reduce the incidence of AD in patients suffering from hypertension and are associated with delay of functional decline in sporadic AD patients (13). Among three subtypes in the ␤AR family, both ␤ 1 AR and ␤ 2 AR play important roles in cognition and stress-dependent behaviors (17,18). Accumulating evidence suggests that ␤ 1 AR and ␤ 2 AR, especially ␤ 2 AR, may be involved in AD pathogenesis through effects on A␤ production or inflammation (14,19,29) and that polymorphisms of ␤ 2 AR contribute to sporadic late-onset AD, which may be related to the availability and response of ␤ 2 AR (13, 16). Our previous studies have shown that A␤ can bind to ␤ 2 AR and induce allosteric activation of the receptor that leads to cAMP/PKA-and GRK/arrestin-mediated cell signaling (19,22,23). In this study, FIGURE 7. The ␤ 2 AR-signaling machinery regulates A␤-induced tau phosphorylation. A␤, amyloid ␤ peptide; ␤ 2 , ␤ 2 receptor; Arr3, arrestin 3. we find that ␤ 2 AR plays a necessary role in A␤-induced tau phosphorylation at Ser-214, Ser-262, and Thr-181 in vitro and in vivo. Deletion of the ␤ 2 AR gene prevents tau hyperphosphorylation, loss of dendritic spines and synapses, and impairment of learning and memory in a transgenic AD animal model. This study places ␤ 2 AR as an essential link between increasing A␤ and tau phosphorylation levels in the brain, which are both hallmarks of AD pathogenesis.
In tauopathies such as AD, frontotemporal dementia, and Parkinson disease, tau is hyperphosphorylated abnormally at multiple serine/threonine sites. In this study, one-year-old PS1/ APP transgenic AD animals show hyperphosphorylation of tau at Ser-214, Ser-262, and Thr-181. Deletion of the ␤ 2 AR gene significantly attenuates the phosphorylation of tau at Thr-181 and completely blocks the phosphorylation of tau at Ser-214 and Ser-262 in vivo and in vitro, suggesting that ␤ 2 AR is a primary receptor for A␤-induced phosphorylation of tau at these sites. PKA and CaMK II are downstream from ␤ 2 AR. Previous studies have shown that both PKA and CaMK II readily phosphorylate tau. However, PKA phosphorylates tau to a significantly greater extent with a broader range of the sites than CaMK II (30,31). Phosphorylation of tau by PKA also significantly decreases tubulin binding (30). In a tandem mass spectrometry study, CaMK II phosphorylated recombinant human tau at the sites, including Ser-214 and Ser-262, that may produce paired helical filament tau (32). Here, we find that A␤-induced phosphorylation of tau at Ser-214 and Ser-262 is primarily dependent on PKA, whereas the phosphorylation at Thr-181 is partially inhibited by PKA inhibitor PKI. These data support that ␤ 2 AR signals through PKA in A␤-induced tau phosphory-lation. In comparison, inhibition of CaMK II does not block the A␤-induced phosphorylation at these sites in PFC neurons. It has been shown that A␤ may induce hyperactivities in AMPA receptors under electric stimulation in PFC slices (22). Here, acute treatment with A␤ alone for 5 min without electric stimulation does not induce significant phosphorylation of CaMK II in both wild-type and ␤ 2 KO PFC neurons, probably because of lack of glutamate released from presynapses for activation of AMPA receptors. Nevertheless, one-year-old PS1/APP transgenic animals show an increased phosphorylation of CaMK II that is dependent on expression of ␤ 2 AR. Thus, a possible role of CaMK II for A␤-induced and ␤ 2 AR-mediated tau phosphorylation in vivo remains to be addressed.
In addition, the JNK pathway amplifies and drives subcellular changes in tau phosphorylation (1) and plays key role in tau phosphorylation in AD models (33). GSK3␤ is a major physiological tau kinase that requires priming phosphorylation at Ser-404 to further phosphorylate tau at paired helical filament 1 (34). In isolated PFC neurons, a JNK inhibitor totally blocks the phosphorylation of tau at Thr-181 and significantly attenuates the phosphorylation of Tau at Ser-214 and Ser-262 induced by A␤. Although A␤ induces phosphorylation of JNK in isolated PFC neurons via a ␤ 2 AR-PKA pathway, the phosphorylation of JNK is only partially blunted in the PFC of 1-year-old ␤ 2 AR-KO/PS1/APP animals. These data indicate that ␤ 2 AR is a major receptor associated with JNK phosphorylation and that other A␤-induced receptor pathways or A␤ deposition-induced inflammation can also promote JNK phosphorylation in vivo. Together, A␤ induces JNK phosphorylation through activating ␤ 2 AR-PKA signaling and other signaling mechanisms in which PKA and JNK independently contribute to tau phosphorylation at Ser-214, Ser-262, and Thr-181.
We also find that A␤ can activate a ␤ 2 AR-arrestin-MAPK pathway in PFC neurons. Surprisingly, we find that MEK, but not downstream ERK1/2 in the MAPK pathway, contributes to phosphorylation of tau at Thr-181. MEK-mediated tau phosphorylation does not require expression of arrestin-2 or arrestin-3, the non-visual arrestins function downstream of ␤ 2 AR in the brain. In comparison, arrestin-3 is required for the A␤-induced ERK1/2 phosphorylation. These data indicate that A␤-induced MEK phosphorylation leads to two divergent pathways: an arrestin-3-dependent ERK1/2 activation and an arrestin-independent tau phosphorylation.
Genetic data have implied that deranged tau-microtubule interactions induced either by phosphorylation or increased levels of tau, contribute to or even are sufficient to cause synaptic and dendritic degeneration in primary tauopathies (35)(36)(37). It has been reported that transfection of tau in mature neurons leads to an improper distribution of tau into the somatodendritic compartment with concomitant degeneration of synapses, as seen by the disappearance of spines and presynaptic and postsynaptic markers (3,7). In this study, there is a degeneration of synapses shown by synapsin I staining and dendritic spines in the deep layer of the PFC of PS1/APP doubletransgenic animals. Deletion of the ␤ 2 AR gene in PS1/APP animals reverses the degenerative effects. These findings again argue for a potential beneficial role of inhibition of ␤ 2 AR in altering the pathological course of AD. Conclusive evidence comes from the behavioral experiments. In the novel object recognition test, learning and memory deficits are present in 1-year-old PS1/APP animals. Deletion of the ␤ 2 AR gene rescues the outcome resulting from overexpressing mutant PS1 and APP genes from human familial AD. It is worth noting that 1-year-old ␤ 2 -KO animals show a tendency to perform better than wild-type animals in the behavioral test. However, 6-month-old ␤ 2 -KO animals show a tendency to have slightly decreased learning and memory. In either case, deletion of the ␤ 2 AR gene in PS1/APP animals has beneficial effects in the test. Taken together, the cellular and behavioral experiments in this study provide evidence that ␤ 2 AR may represent a potential therapeutic target for preventing the development of AD.