LiCl attenuates impaired learning and memory of APP/PS1 mice, which in mechanism involves α7 nAChRs and Wnt/β‐catenin pathway

Abstract We examined the mechanism by which lithium chloride (LiCl) attenuates the impaired learning capability and memory function of dual‐transgenic APP/PS1 mice. Six‐ or 12‐month‐old APP/PS1 and wild‐type (WT) mice were randomized into four groups, namely WT, WT+Li (100 mg LiCl/kg body weight, gavage once daily), APP/PS1 and APP/PS1+Li. Primary rat hippocampal neurons were exposed to β‐amyloid peptide oligomers (AβOs), LiCl and/or XAV939 (inhibitor of Wnt/β‐catenin) or transfected with small interfering RNA against the β‐catenin gene. In the cerebral zone of APP/PS1 mice, the level of Aβ was increased and those of α7 nicotinic acetylcholine receptors (nAChR), phosphor‐GSK3β (ser9), β‐catenin and cyclin D1 (protein and/or mRNA levels) reduced. Two‐month treatment with LiCl at ages of 4 or 10 months weakened all of these effects. Similar expression variations were observed for these proteins in primary neurons exposed to AβOs, and these effects were attenuated by LiCl and aggravated by XAV939. Inhibition of β‐catenin expression lowered the level of α7 nAChR protein in these cells. LiCl attenuates the impaired learning capability and memory function of APP/PS1 mice via a mechanism that might involve elevation of the level of α7 nAChR as a result of altered Wnt/β‐catenin signalling.

of neurons participate in cognitive processing, as well as learning and memory. Expression alterations in these receptors are associated with the impaired cholinergic neurotransmission 4 and may be responsible for complex neuronal diseases such as AD and Parkinson disease. 5 Indeed, nAChRs level is lowered in the cerebral zone of the AD patients and of animal models of this disease, as well as in isolated neuronal cells exposed to Aβ. 6,7 α7, the main subtype of nAChR present in both neuronal and non-neuronal cells of the human brain consists mainly of five α7 subunits and is considered to be one of the receptors most closely associated with the AD contraction. 8 Aβ  and Aβ  bind to α7 nAChR with an affinity in the nanomolar range, resulting in gradual depletion and inactivation of this receptor. 9 In addition to the toxic consequences due to the direct nAChRs binding by Aβ, other subcellular events have been associated with AD, including down-regulation of canonical axis Wnt/β-catenin, which might influence the stability of these receptors. When Wnt receptors are not activated, the β-catenin bound by Axin is phosphorylated in succession via glycogen synthase kinase-3β (GSK3β) and casein kinase 1 at a range of Ser/Thr residues (regularly spaced), which are located at the N-terminus of the β-catenin. Consequently, ubiquitination and targeting of β-catenin are initiated for fast disruption by proteasome, thereby depressing the expression of downstream targets such as cyclin D1. 10 Interestingly, Wnt/β-catenin signalling is closely linked to learning capability and memory function among mammals and this signalling is impaired both in the brains of AD patients, 11 as well as animal AD models. 12 This dysfunction exerts a crucial effect on the neuronal degeneration, as well as the impairment of synapses, creating a deleterious pathway that leads to dementia. 13 Notably, nAChRs are present both on presynaptic 14 and on postsynaptic 15 membranes, and loss of these receptors leads inevitably to abnormalities in synapses. Furthermore, when the Wnt signalling pathway is activated persistently via ligand medication or suppressor deactivation, it overcomes the toxic effects of Aβ and improves cognitive performance in patients with AD. 16,17 As is widely acknowledged, lithium is the preferred moodstabilizing agent for maintaining bipolar disorder. 18 Apart from stabilizing moods, lithium can also resist suicidal thoughts, regulate immunity and protect nerve system, 19 which, during the last decade, has come to be regarded as a neuroprotective agent and is now widely used in studies of neurodegenerative diseases, including AD. 20 It is noteworthy in this context that the incidence of AD among patients with bipolar disorder who have been taking lithium is lower than among those not undergoing lithium therapy. 18 Lithium chloride (LiCl) inhibits GSK3β, thereby reducing this activity in the brain of animal models of AD, while enhancing Wnt/β-catenin signalling. 21,22 Accordingly, lithium may be of value in treating AD. However, it is currently unknown whether lithium can reverse the learning and memory impairments in murine AD model and, if so, whether this effect involves Wnt/β-catenin signalling modulation, as well as consequent regulation of α7 nAChR. Here, we investigated these questions by treating both APP/PS1 transgenic mice and β-amyloid peptide oligomer (AβO)-exposed primary neurons using LiCl or XAV939, a small molecule selective inhibitor that promotes β-catenin degradation and inhibits the transcription by β-catenin, 23 and then analysing the cognitive ability of these mice, as well as potential changes in Wnt/βcatenin and α7 nAChR in both of these systems.

| Experimental animals
B6.Cg-Tg (PSEN1dE9 and APP SWE ) mice aged 4 months having a 85Dbo/Mmjax background were procured from Nanfang BioTechnology Co., Ltd. Meanwhile, wild-type (WT) mice of identical strain were also procured from the same company. All mice were 20-30 g in body weight. Each male mouse was allowed to mate with four females. When the resulting pups were 12-20 days old, the tips of their tails were cut off for extraction of DNA and the polymerase chain reaction (PCR)-based genotyping, where agarose gel electrophoresis (1.5%) was used for product analysis. For PCR primer design of target transcripts shown in Table 1, the entire cDNA sequences from GenBank were consulted.

| Culturing and treatment of primary hippocampal neurons
Primary neurons were obtained from the cerebral zone of newborn Sprague-Dawley rats employing minor modifications of a published procedure. 26 For purity assessment of these primary neurons, immunofluorescent double staining was performed in succession using the anti-NeuN mouse antibody, the CY-3-labelled anti-mouse IgG (red), the anti-GFAP rabbit antibody and a subsequent 488-labelled anti-rabbit IgG (green).
AβOs were prepared by a procedure described earlier 27 and the suitable concentration for exposure chosen as also described previously. 26 Initially, cell exposure to varying doses of LiCl (0-100 mmol/L) and XAV939 (0-50 μmol/L) was accomplished for different times.
Thereafter, the cells were subjected to 2-h incubation with CCK8 solution (10 μl), following which the absorption at 450 nm was determined.
Transfection with small interfering RNA (siRNA) was ini-

| Morris water maze for spatial learning and memory
In a 25-26°C water-filled round pool rendered opaque using milk powder, every mouse was compelled to find an underwater escape facility. 29 In the familiarization and acquisition sessions, 60 s was offered for every mouse to seek the hidden facility, and then, 5 s was offered to keep seated on the facility. Afterwards, the mice returned to the original cage. In the retention session, where the facility was withdrawn from the pool, the path adopted by every mouse was video recorded for 60 s, in order to determine the time expenditure for swimming to the original facility location, and the number of passes across and the time expenditure at this location.

| Determination of Aβ 42 by ELISA
The murine cerebral level of Aβ 42 was determined utilizing a murine To achieve relative protein quantization, the above PVDF membranes were then subjected to incubation using antibodies against phosphor-GSK3β (ser9), GSK3β, β-catenin, cyclin D1, α7 nAChR or GAPDH overnight at 4°C. Next, the membranes were rinsed and subjected to 60-min incubation using secondary antibody conjugated to HRP. Subsequently, enhanced chemiluminescence kit (Millipore) was used to detect the protein bands, and visualization of the resulting signals was accomplished for 30 s-3 min by exposing to chemiluminescence film (hyperperformance). Signal intensity was quantified employing the Image J software.

| Determination of the level of α7 nAChR mRNA by reverse transcription and quantitative realtime PCR
Total RNA, which was extracted via Trizol Reagent from brain tissue or cells, was used to obtain cDNA by reverse transcrip-  Table 2. The level of GAPDH mRNA was used as the internal control.

| Examination of senile plagues and α7 nAChR by immunohistochemical or immunofluorescent staining
Immunohistochemical staining for senile plagues and α7 nAChR in the brains of mice was accomplished by consulting prior literature. 31 Initially, sections were deparaffinized and dehydrated. Next, they were subjected to 20-min microwaving with pH 6. For immunostaining of α7 nAChR in isolated cells, seeding of primary neurons was done onto six-well plates coated with polylysine, followed by three times of washing in PBS and paraformaldehyde fixation (4%). Then, these samples were subjected to 30-min treatment with goat serum at RT and a subsequent overnight incubation with antibody against α7 nAChR at 4°C. 26

| Statistical analyses
For mice in various groups, the statistical values were represented as means ± SDs and compared via analysis of variance, and a subsequent post hoc (least significant difference) test.
Bielschowsky procedure was employed to semi-quantitate the Aβ plaques. 32 The correlation of the level of α7 nAChR protein with spatial learning and memory was analysed employing the

| Confirmation of the APP/PS1 murine genotype
As shown in Figure S1, when the PCR amplification template was genomic DNA, the transgenic group exhibited products (sizes around 400 and 600 bp) that coincided separately with the APP and PS1 genes in size. However, this was not observed among the WT mice.

| Spatial learning and memory in mice
According to the Morris water maze results displayed in Figure 1 time expenditure at this location were reduced, while the escape latency was increased in the 6-and 12-month-old APP/PS1 mice as compared to the corresponding WT mice. Treatment for the transgenic animals with LiCl for 2 months beginning at 4 or 10 months of age clearly improved both their learning capability and memory function.

| Senile plaque count and level of Aβ 42 in the cortex of mouse brains
The brains of 6- (Figure 2A-D and a-d) or 12-month-old ( Figure 2E-H and e-h) transgenic mice contained Aβ-immunoreactive senile plaques, the number of which was markedly reduced by exposure to LiCl beginning at either 6 or 12 months of age. There were no detectable amyloid plaques in the cerebral zone of WT mice with or without exposure LiCl ( Figure 2I).
The cortex of WT mice contained a very low level of Aβ 42 that was unaffected by treatment with LiCl. In contrast, this level was much higher in untreated APP/PS1 animals, where the level was decreased even further by treatment with LiCl ( Figure 2J).

| Primary neuronal viability upon AβOs, LiCl and/or XAV939 exposures
As shown in Figure S2A-E, the immunostaining analysis of the primary neurons cultured using antibody, which were prepared from the cerebral hippocampal zone of neonatal rats, directed towards DAPI (a nucleus marker), GFAP (an astrocyte marker) and NeuN (a neuron marker), suggesting that these cells (about 90%) were neuronal. These primary neurons were assessed for viability following Treatment with 10 mmol/L LiCl for 6 h or 1 μmol/L XAV939 for 24 h did not cause any significant cytotoxicity ( Figure S2F,G). Exposure of these primary neurons to 0.5 μmol/L AβOs for 48 h (conditions chosen on the basis of an earlier study; 33 ) reduced cell viability, a reduction that was attenuated by LiCl, but aggravated by XAV939 ( Figure S2H).

| Expression of proteins involved in Wnt signalling in the hippocampus and cortex of mouse brains and primary neurons
As Western blotting revealed, the phosphor-GSK3β (ser9) (not total Noteworthy is that such declines were mitigated by LiCl treatment for transgenic mice. With regard to primary neurons, the levels of the β-catenin and cyclin D1 proteins were enhanced by exposure to LiCl alone, whereas XAV939 reduced these levels. Exposure of primary neurons to AβOs alone also reduced these levels, an effect that was attenuated by LiCl and enhanced by XAV939 ( Figure 4A,B).
Compared to the corresponding controls, transfection of primary neurons with siRNA targeting the β-catenin gene reduced both the level of this protein and that of cyclin D1 ( Figure 4E,F).

| Expression of α7 nAChR mRNA and protein in the hippocampus and cortex of mouse brains and primary neurons
In comparison with the corresponding control group, the levels of both α7 nAChR protein and mRNA were enhanced by

F I G U R E 4
Expression of the β-catenin (A and E) and cyclin D1 proteins (B and F) and α7 nAChR protein and mRNA (C, D and G) by primary hippocampus neurons. control: untreated primary neurons; Li: primary neurons exposed to LiCl (10 mmol/L) for 6 h; XAV939: primary neurons exposed to XAV939 (1 μmol/L) for 24 h; Aβ: primary neurons exposed to AβOs (0.5 μmol) for 48 h; Li+Aβ: primary neurons exposed to LiCl and AβOs; XAV939+Aβ: primary neurons exposed to XAV939 and AβOs; sicontrol: primary neurons transfected with sicontrol; siβ-catenin: primary neurons transfected with small interfering RNA (siRNA) targeting the β-catenin gene for 48 h. The values presented are the means ± SDs from three separate experiments. *p < 0.05 as compared to control group; # p < 0.05 as compared to the Aβ group exposure of the neurons to LiCl alone and reduced by XAV939.
Moreover, these levels were reduced by exposure to AβOs, a decline that was attenuated by LiCl, but enhanced by XAV939 ( Figure 4C,D).
Moreover, transfection of neurons with siRNA targeting the βcatenin gene reduced the level of this protein, as well as that of α7 nAChR ( Figure 4G).
According to the results of real-time PCR and Western blotting, declines in the protein ( Figure 5A-D) and mRNA ( Figure 5E-H) levels of α7 nAChR were noted in the hippocampus ( Figure 5A,C,E,G) and cortex ( Figure 5B,D,F,H) of APP/PS1 mice at 6 ( Figure 5A,B,E,F) and 12 ( Figure 5C,D,G,H) months of age. Administration of LiCl to either the WT or APP/PS1 animals elevated these levels in both of these areas of the brain.
Semiquantitative immunohistochemical analysis of mouse brains revealed localization of α7 nAChR primarily at the plasma membrane and axon ( Figure 6A-D and a-d). This immunostaining was less intense in the hippocampus ( Figure 6A,a,C,c) and cortex ( Figure 6B,b,D,d) of the APP/PS1 mice at 6 ( Figure 6A,a,B,b) and 12 ( Figure 6C,c,D,d) months of age than for the WT group.
In both groups of animals, LiCl augmented the intensity of this staining.
In the case of primary neurons, semiquantitative immunofluorescent analysis revealed localization of the α7 nAChR subunit mainly at the plasma membrane and axon ( Figure 7A). Exposure to LiCl alone enhanced this immunofluorescence, whereas XAV939 reduced it.
Interestingly, the reduction caused by AβOs alone was attenuated by LiCl and enhanced by XAV939 ( Figure 7D).

| Correlations between the α7 nAChR level and spatial learning and memory
Correlation analysis revealed a negative association between the elevated level of α7 nAChR protein present in the cerebral zone of APP/PS1 mice aged 6 ( Figure S3A,C) or 12 ( Figure S3B,D) months following exposure to LiCl and the impairment in their spatial learning and memory.

| DISCUSS ION
In the cerebral zone of APP/PS1 mice, one most common animal AD model, the content of Aβ and number of senile plaques are elevated, while the learning capability and memory function of these mice are impaired. 34 Here, we found that in the cerebral zone of 6-and, especially, 12-month-old mice of this strain, the increases in size and quantity of senile plaques were evident and learning and memory significantly impaired. Consistent with previous studies, 25,26 we have also confirmed that Aβ damages neurons directly, both in vitro and in vivo, leading to dysfunction and apoptosis.
α7 nAChR is widespread throughout the central nervous system.
Regarding the well-defined intracerebral functions of α7 nAChR, it can modulate synaptic plasticity and transmission that underlay ordinary processes of cognition, attention, learning and memory. 35,36 In our current investigation, the protein expression of α7 nAChR in the cerebral hippocampus and cortex of APP/PS1 mice aged 6 and 12 months was found to be lower than in WT animals. In addition, The values are represented as means ± SDs (n = 6), *p < 0.05 as compared to WT; # p < 0.05 as compared to APP/PS1 mice decline of learning and memory in patients with AD also appears to be associated with the decrease of Wnt/β-catenin in their brain. 39 In animal models as well, the AD pathology is strongly correlated with the down-regulation of Wnt/β-catenin axis. 12 Accordingly, the Wnt/β-catenin signal pathway offers a therapeutic target for alleviating the pathological process that results in AD. 40 In the present case, we verified that Wnt/β-catenin signalling in the APP/PS1 murine cerebral zone or in the primary neurons exposed to AβOs in vitro is evidently weaker than in the WT mice or unexposed cells.
Evidence for a neuroprotective effect of lithium has accumulated over the last two decades, 41 and this element has been found to be involved in the regulation of numerous genes, proteins and metabolites. 42 Most noteworthy of these is the inhibition of GSK3β by lithium, which has a wide range of beneficial effects. 43 Consistent with previous findings, 44 we found here that LiCl enhances the learning capability and memory function of APP/PS1 mice aged both 6 and 12 months.
Moreover, both the elevated content of Aβ 42 and the size or quantity of senile plaques in the cerebral zone of these transgenic mice were reduced through treatment with LiCl. In addition, this reduction in Aβ content was correlated to an improvement in learning and memory. Some evidence from animal models of AD indicates that LiCl can prevent neurotoxicity by reducing the production of Aβ, 45 as well as that the cognitive benefits of LiCl are associated with enhanced clearance of Aβ via up-regulation of brain microvascularization by lipoprotein receptor-related protein 1 and increased bulk flow of cerebrospinal fluid. 46 In human neuronal cells, LiCl alters proteins such as the rab proteins, which have been implicated in the processing of APP in connection with the pathophysiology of AD. 47 In addition, upregulation of Wnt/β-catenin signalling by LiCl can cause β-catenin to bind specifically to regions within the BACE1 promoter that contain putative TCF/LEF motifs and repress transcription. 33 Any of these mechanisms may explain the reduction in Aβ caused by lithium.
LiCl is a classic agonist of the Wnt/β-catenin pathway, which is the main reason it has been chosen for testing as a candidate drug for treating AD in numerous studies. 10,22 Here, up-regulation of the Wnt/β-catenin axis by LiCl in 6-and 12-month-old mice with or without the APP/PS1 double mutation was verified. This effect was associated with inactivation of GSK3β through specific phosphorylation of Ser9 in its N-terminus, 48 which elevates the levels of the β-catenin and cyclin D1 proteins. In addition, we found here that in primary hippocampal neurons the Wnt/β-catenin pathway was activated by LiCl and inactivated by XAV939 or AβOs. Importantly, this effect by AβOs could be attenuated by LiCl, but aggravated by XAV939.

F I G U R E 7
Immunofluorescent staining for α7 nAChR protein in primary hippocampus neurons. Control group: untreated primary neurons; Li: primary neurons exposed to LiCl (10 mmol/L) for 6 h; XAV939: primary neurons exposed to XAV939 (1 μmol/L) for 24 h; Aβ: primary neurons exposed to AβOs (0.5 μmol) for 48 h; Li+Aβ group: primary neurons exposed to LiCl and AβOs; XAV939+Aβ group: primary neurons exposed to XAV939 and AβOs. The neurons were stained using rabbit antibody against α7 nAChR (green), while staining of cellular nuclei was accomplished using DAPI (blue). D: semiquantitative analysis of the integrated optical density (IOD) for α7 nAChR staining. Magnification: 400×, scale bar = 25 μm. The values are represented as the means ± SDs from three separate experiments. *p < 0.05 as compared to the control group; # p < 0.05 as compared to the Aβ group In line with these findings in vitro, the Wnt/β-catenin axis initiation has been proven to offset the harmful action of AβO, 40 which thus exerts a probable fundamental effect on maintaining neuronal and synaptic functions. 49 Abundant evidence demonstrates that reactivation of the Wnt/β-catenin pathway completely restores the number of synapses and synaptic plasticity 50 and this pathway is a target for recovery of neuronal circuits following degeneration of synapses. 51 Since α7 nAChRs are expressed widely on both presynaptic and post-synaptic membranes, 14,15 we hypothesize that activation of Wnt/β-catenin pathway by LiCl stabilizes expression of these receptors in the animal AD models' cerebral region.
Interestingly, the reduction in the cerebral α7 nAChR protein ex- Interestingly, a possible role for the expression and activity of N-methyld-aspartic acid receptors (NMDARs) and glutamatergic transmission was found in the therapeutic effects of lithium. 57 In many cases, the α7 nAChRs and NMDARs were closely apposed on axon terminals, where α7 nAChRs may regulate glutamate release and its subsequent modulation of neuronal plasticity. 58,59 Activating α7 nAChRs potentiated responses to glutamate by increasing the membrane insertion of NMDARs. 60 However, whether lithium affects α7 nAChRs signal transduction through NMDARs has not been verified, which is worth of further study.
In the cerebral zone of APP/PS1 mice, the Aβ level was increased and those of α7 nAChR, phosphor-GSK3β (ser9), β-catenin and cyclin D1 (protein and/or mRNA levels) were reduced. Treatment with LiCl for 2 months at 4 or 10 months of age attenuated all of these effects.
Similar expression variations were observed for these proteins in primary neurons exposed to AβOs, and these effects were attenuated by LiCl and aggravated by XAV939. Inhibition of β-catenin expression lowered the level of α7 nAChR protein in these cells. A proposed molecular pathway involving above changes is shown in Figure S4.
These findings suggest that LiCl can reverse the impairment in learning capability and memory function for APP/PS1 mice via a mechanism that might involve elevation of the level of α7 nAChR as a result of altered Wnt/β-catenin signalling.

CO N FLI C T O F I NTE R E S T
The authors have no conflict of interest to declare.

DATA AVA I L A B I L I T Y S TAT E M E N T
All data generated or analysed during this study are included in this published article and its supplementary information files.