Endoplasmic reticulum stress induces spatial memory deficits by activating GSK‐3

Abstract Endoplasmic reticulum (ER) stress is involved in Alzheimer's disease (AD), but the mechanism is not fully understood. Here, we injected tunicamycin (TM), a recognized ER stress inducer, into the brain ventricle of Sprague‐Dawley (SD) rats to induce the unfolded protein response (UPR), demonstrated by the enhanced phosphorylation of pancreatic ER kinase (PERK), inositol‐requiring enzyme‐1 (IRE‐1) and activating transcription factor‐6 (ATF‐6). We observed that UPR induced spatial memory deficits and impairments of synaptic plasticity in the rats. After TM treatment, GSK‐3β was activated and phosphorylation of cAMP response element binding protein at Ser129 (pS129‐CREB) was increased with an increased nuclear co‐localization of pY126‐GSK‐3β and pS129‐CREB. Simultaneous inhibition of GSK‐3β by hippocampal infusion of SB216763 (SB) attenuated TM‐induced UPR and spatial memory impairment with restoration of pS129‐CREB and synaptic plasticity. We concluded that UPR induces AD‐like spatial memory deficits with mechanisms involving GSK‐3β/pS129‐CREB pathway.


| INTRODUCTION
Alzheimer's disease (AD) is a chronic neurodegenerative disorder with progressive impairment of memory and other cognitive functions. Apart from the loss of synapses, AD is histopathologically characterized by the accumulation of numerous intracellular neurofibrillary tangles (NFTs) and extracellular plaques. 1,2 Previous studies have shown that the amount of NFTs, mainly composed of the hyperphosphorylated tau, 3,4 in the AD brains is correlated with the degree of dementia. 5,6 Studies also showed that b-amyloid (Ab), forming toxic oligomers that aggregate into amyloid plaques, was associated with age-related memory impairment. 7,8 Our previous studies indicate that tunicamycin (TM) could induce AD-like tau hyperphosphorylation and reduction in some synapse-related proteins in temporal cortex, frontal cortex and hippocampus. 9 However, whether TM treatment affects learning and memory and the molecular mechanisms are unknown.
Endoplasmic reticulum (ER) is an important cellular organelle, responsible for the posttranslational processing of newly synthesized proteins and ensuring proper protein folding and assembly. 10 ER stress is an important form of ER dysfunction, and ER stress has been observed in several neurological conditions, such as AD, Parkinson's disease, Amyotrophic lateral sclerosis and so on. Some studies have showed neurons are constantly exposed to ER stress in the AD brains. ER stress could be expressed by chaperone proteins and trigger many rescuer responses, including unfolded protein response (UPR) and ER-associated degradation. [11][12][13] The ER chaperone binding immunoglobulin protein (Bip) is physiologically bound to 3 important proteins in the ER membrane, pancreatic ER kinase (PERK), inositol-requiring enzyme-1 (IRE-1) and activating transcription factor-6 (ATF-6). When UPR is induced, Bip is attracted to bind to the unfolded proteins accumulated in the ER to keep the correct protein folding and is thereby released from PERK, IRE-1 and ATF-6, which are consequently phosphorylated and activated. [14][15][16] Although the initial UPR protects the cell from the toxicity of misfolded proteins in the ER, prolonged UPR activation may participate in the pathogenesis of protein misfolding diseases, such as AD. [17][18][19][20] Recently, several reports have shown that the UPR is activated in the AD brain. Bip, an ER stress marker, is increased in the temporal cortex and the hippocampus of AD cases compared with no demented control cases. 21 The phosphorylated PERK (pPERK), an UPR activation marker, is most abundant in neurons with diffuse localization of the phosphorylated tau protein in the brain of AD patients. 22 Our previous report has also shown that TM treatment induces tau hyperphosphorylation in frontal cortex, temporal cortex and hippocampus in rats with an increased level of Bip and reduction in some synapse-related proteins, 9 while overexpressing SIL1 rescued Bip elevation-related Tau hyperphosphorylation in ER stress. 23 Glycogen synthase kinase-3 (GSK-3) is highly expressed in the central nervous system (CNS) 24 and plays an important role in AD. Studies showed that the activated form of GSK-3 was elevated in the AD brains. 25 GSK-3 could not only phosphorylate tau at most of the AD sites [26][27][28] but also induce Ab overproduction. 29 Activation or overexpression of GSK-3 induces memory deficit, [30][31][32] whereas inhibition of GSK-3 reverses this effect. 31,33 However, the mechanism by which GSK-3 regulates learning and memory is only partly understood. Recent study has found that overexpression of GSK-3b could cause memory deficits by inhibiting long-term potentiation which is accompanied by prominent impairment of synapses. 34,35 An in vitro study also show that GSK-3b could be activated during ER stress 36,37 and induce tau hyperphosphorylation 23,38 that be involved in memory impairment. However, it is still not understood whether and how GSK-3b plays an in vivo role in ER stress-induced spatial cognitive alterations.
The cAMP response element binding protein (CREB), named by Montminy, 39 regulates transcription of multiple genes in eukaryotic nuclei. CREB plays an important role in increasing long-term potentiation (LTP), synaptic plasticity, development, differentiation and survival of neurons. Phosphorylation of CREB at Ser133 increases CREB activity, whereas phosphorylation at Ser129 and Ser142 inhibits its activity. GSK-3b is an important kinase regulating the transcription of CREB. 40 In this study, we established a rat model with activated UPR by brain injection of TM. We found that TM infusion induces spatial memory deficits in rats with ER stresses, shown by the increased level of phosphorylated PERK, IRE-1, ATF6, CREB at Ser129, GSK-3b at Tyr216 and impairment of synapses. Simultaneous inhibition of GSK-3 rescues the UPR-induced spatial memory impairments with restoration of ER stress and the associated dysfunction.

| Antibodies and chemicals
The antibodies used in this study are listed in Table 1  The rats (rats for each group were used in this study are listed in Table 2) were anaesthetized with 6% chloral hydrate (400 mg/kg) and placed in a Jiangwan-II stereotaxic instrument (Jiangwan Medical Instrument Co. Shanghai, China). 41 The skull was cleaned, and the hole (diameter 1.0 mm) was made for the infusion after the scalp was incised (5.0-8.0 mm). For the lateral ventricular infusion, the coordinate of AP-0.8, L-1.5, V-4.0 (in mm from bregma and dura, flat skull) was selected according to the stereotaxic atlas of Franklin and Paxinos. A sterilized needle connected to a Hamilton syringe was used to deliver TM or in combination with SB into the lateral ventricle (10 lL). Equal volume of DMSO with 0.9% NaCl was infused as vehicle controls.

| Behavioural test
Spatial memory was measured by Morris water maze test. 42 The Morris water maze apparatus was the same as previously described. 43 In brief, it consisted of a circular pool, 150 cm in diameter and 50 cm in height, with the interior painted black. The escape platform was made of clear plexiglass, 14 cm in diameter and 27 cm in height and was located 1 cm below the surface of the water.
Visual cues were visible to the rats, including several geometric shapes that measured at least 20 cm in height and were positioned so that they were 15 cm above water level and remained distal and constant to the rats at all times. The water was maintained at 24 AE 2°C and was made opaque by the addition of non-toxic black ink that generated the obvious contrast with the white skin of rats to record their movements. The swimming pathways and the latencies of the rats to find the hidden platform were recorded each day by a video camera fixed to the ceiling of the room, 1.5 m from the water surface. The camera was connected to a digital-tracking device attached to an IBM computer loaded with the water maze software.
The less time a rat spent in finding the platform, the better it scored the spatial learning and memory.
For spatial learning, rats were trained in water maze to remember the hidden platform for 7 consecutive days. For the first day, the rats only proceeded with one trial that started from the first quadrant. And for the following 5 days, 3 trials were performed for each rat every day, starting from the first, the second and the fourth quadrant, respectively. For the last training day (the 7 day), only one trial that started from the first quadrant's movements were performed, and total 20 trials were finished for 7 days' training. On each trial, the rat started from each quadrant by facing the wall of the pool and ended when the animal climbed on the platform. The rats were not allowed to search for the platform more than 60 seconds, after which they were guided to the platform. Through these training sessions, rats acquired spatial memory about location of the safe platform, and rats that could find the platform after training from quadrant 1-4 in turn for 24 trials within 20 seconds were selected and randomly divided into 3 groups for the brain lateral ventricle injections, respectively, with DMSO (10 lL) or TM T A B L E 1 A list of antibodies and their epitopes on the molecule of protein used in this study  Put the rats into the experimental device to make them be familiar with the surrounding environment for 5 minutes, respectively. And then began the stage of learning to put the rats gently on the platform, when animals jumping off the platform to the wire fence, immediately gave them electric shock (0.5 mA, for 5 seconds). After that put them back to the platform and began to record duration, if the duration reached 60 seconds that indicated rats have learned how to avoid electric shock. If the duration on the platform was <60 seconds, gave the rats electric shock again when they jumped off the platform until they could stay on the platform at least 60 seconds. Record the number of rats received shocks (mistake number) and the total time (learning time for the first time), the whole period was 5 minutes. The laboratory equipment should be cleaned with 75% alcohol after each rat was tested. When all rats were trained, following start the formal test to check the memory retention at 24 and 48 hours after DMSO or TM being injected. The rats were put on platform in this phase experiment, and the duration period and numbers of wrong were recorded within 5 minutes.

| Cytoplasm-nucleus protein extraction
The 250-mg fresh rats brain or cryopreserved tissue in À80°C were taken immediately into the glass homogenizer with ice pre-cooling, add 500 lL cytosol extraction reagent (CER, the proportion of brain tissue with CER is 1:2) in it. Triturate the brain tissue with pestle   Table 1 for detail) at 4°for

| Western blotting
overnight. Then, the blots were incubated with anti-mouse or antirabbit IgG conjugated to horseradish peroxidase (1:15 000) for 1 hour at room temperature and scanned after being washed with TBS-Tween-20,and the greyscale was analysed with odyssey system.

| Nissl staining
Nissl staining was established in 1892 by Franna Nissl, the German pathologist, with alkaline dye to discover Nissl body, and this method was widely used. The picked brain slices were pasted on the   Figure 1E,F). In our previous study, we observed that TM could activate GSK-3b. Therefore, we studied whether simultaneous inhibition of GSK-3 by SB216763 (SB) affects UPR. The results showed that application of SB did not rescue UPR ( Figure 1E,F). These results suggest that ventricular infusion of TM can induce UPR in rat brain independent of GSK-3 activation.
In our previous studies, we found that TM treatment increased phosphorylated level of tau at Thr205, Thr231 and Ser396. 9 We observed the similar alternation of tau proteins in this study and SB attenuated tau phosphorylation (Figure 2A,B).
To measure the effects of UPR in spatial memory, we trained the rats for 7 consecutive days to allow remembering the hidden platform in water maze ( Figure 2C

| TM inhibits mushroom spine formation and expression of several synaptic proteins
To explore the mechanisms underlying the TM-induced spatial memory deficits, we measured spine morphology and synapse-associated proteins. We found that number of mushroom-type spines therefore, we measured GSK-3b using activity-dependent antibodies.
We found that the active phosphorylation of GSK-3b at Tyr216 (pY216-GSK-3b) increased and the inhibitory phosphorylation of F I G U R E 2 SB attenuates tau hyperphosphorylation and memory deficits induced by tunicamycin in rats. The rats were randomly divided into 3 groups infused, respectively, through ventricle with 50 lmol/L TM or DMSO or TM plus SB (50 lmol/L). The rats were trained in Morris water maze (MWM) for 7 days before DMSO, TM and TM + SB injection (C). After 24 or 48 h, the brain extract from hippocampal regions (HP) was used to measure the alterations of tau proteins by Western blotting (A) and quantitative analysis (B). The levels of unphosphorylated tau at Tau1 epitope and the phosphorylation level of tau at Ser396 epitope as labelled on the blot were normalized against total tau probed by Tau5 which was normalized against DM1A (n = 6). SB could more obviously rescue the decreased Tau1 and the increased phosphorylation level of tau at Ser396 epitope after TM being injected for 48 h. Simultaneously, the MWM and step-down avoidance tests were used to assess learning and memory capacities (D-I). The rats had same cognitive levels during 7 days training before TM treatment (C), while injection of TM for 24 or 48 h induced memory deficits shown by the increased latency to find the hidden platform in MWM test (D,E). TM-injected rats used more time to learn to protect themselves from the risk of electric shock in the training period during step-down avoidance test measured at 24 and 48 h after the injection (F). TM-injected rats showed no difference of the number of errors compared with the control group rats (G). TM-injected rats showed no difference of latency period but increased number of errors in the detection period during step-down avoidance test measured at 24 and 48 h after the injection. SB rescued TM-induced memory deficits shown by the significantly decreased latency to find the hidden platform in MWM test (E). The data were expressed as mean AE SD (n = 10). *P < .05, **P < .01 vs DMSO-24 h; # P < .05, ## P < .01 vs DMSO-48 h; ■ P < .05vs TM-24 h, ▲ P < .05 vs TM-48 h in B; *P < .05, **P < .01 vs DMSO; # P < .05 vs TM in E-I F I G U R E 3 Tunicamycin inhibits mushroom spine formation and expression of synaptic protein and attenuation by SB. The representative images of dendritic spines in rat hippocampal CA1, CA3 and DG at 48 h after TM injection (A,B). TM decreased numbers of mushroom-type spines significantly in DG and CA3 subsets but not in CA1 of the TM-treated group, and no significant change of thin-spines was detected (B-F). SB could reverse the above phenomenon. The levels of synapse-associated proteins were measured by Western blotting and quantitative analysis, normalized against tubulin probed by DM1A (G,H). The data were expressed as mean AE SD (n = 3 for A and B, bar = 50 lm for A, bar = 2 lm for B; n = 6 for G). *P < .05, **P < .05 vs DMSO, # P < .05 vs TM

| DISCUSSION
In the AD brains, the immunoreactivity of the ER stress markers, such as pPERK, eIF2a and IRE-1a, was observed in hippocampal neurons associated with granulovacuolar degeneration, and the pPERK-immunoreactive neurons were increased. 21  . SB treatment restored the nuclear internal and external metastasis of pSer133-CREB and pSer129-CREB. The co-localization of pY216-GSK-3b with p-S129-CREB and p-S133-CREB in hippocampus were statistically measured. The data were expressed as means AE SD. The tendency of pY216-GSK-3b with p-S129-CREB and p-S133-CREB in cortex was similar that in hippocampus (F,G) and the statistical graph not be showed. (n = 6) *P < .05, **P < .01, ***P < .001 vs DMSO in B and C. *P < .05 vs DMSO, # P < .05 vs TM in F and G proteins. 22  were more vulnerable to the stresses. 50 Meanwhile, the dendritic spines in DG and CA3 subsets also decreased accompanied with the mushroom type of dendritic spines of hippocampus. The mushroom type of dendritic spines has been reported to be closely related to memory, and its plastic was limited. Moreover, the postsynaptic associated proteins, PSD95 and synapsin 1 significantly decreased after TM injection and SB could rescue the decrease in PSD95 but not synapsin 1; furthermore, it decreased the level of synapsin 1. The mechanism may deserve further investigation.
GSK-3b is a key kinase that plays a crucial role in AD-like tau hyperphosphorylation. 51,52 GSK-3b activation or conditionally overexpressed GSK-3b has been previously reported to cause spatial memory deficits in animals and inhibiting GSK-3b could revise AD-like cognitive deficit. 43,53,54 An in vitro study also shows that GSK-3b is activated during ER stress. 38 However, it is still not understood whether and how GSK-3b plays an in vivo role in ER stress-induced tau phosphorylation and cognitive alterations. In this and our previous studies, we showed that UPR induced by TM and activated GSK-3b that resulted in tau hyperphosphorylation in vivo and impaired spatial memory of rats. Inhibiting GSK-3b by GSK-3b inhibitor SB216763 could reverse the spatial memory detentions induced by ventricle brain injection TM.
In addition, studies proved that GSK-3b could mediate phosphorylation of CREB, but its function and the mechanism was still not clear. Jason L. et al found that inhibitor of GSK-3 could reduce the expression of fluorescein of RAW-CRE cell treated with water toxin, which mediated by CREB. Simultaneously, they found water toxin did not activate the level of PKA-dependent phosphorylation of CREB at Ser133 but increased the level of phosphorylation of CREB at Ser129. 55 Some studies have shown that GSK-3b could make CREB phosphorylate at Ser129. But there were conflicting views on CREB phosphorylation at Ser129. Some believe that phosphorylation at Ser129 causes the trans-activation of GAL4-CREB fusion protein and then promote the expression of syntrophic transcription factor and gene. 56,57 Some others reported that activating GSK-3 and/or increasing the level of pS129-CREB inhibit the transcription activity of CREB by decreasing the bonding affinity of CREB and DNA. [58][59][60] In the present study, we found that TM could active GSK-3b and increase the level of CREB phosphorylation at Ser129 in hippocampus, consequently resulted in a fall in learning and memory ability relevant CREB. At the same time, we found that TM could induce the increasing CREB phosphorylation at Ser129 locus in nucleus and CREB phosphorylation at Ser133 locus in cytoplasm of cortex and hippocampus, which maybe relate to CREB transfer both inside and outside the nuclear induced by TM and then regulate its downstream target genes.
Taken together, we find in the present study that TM-induced UPR causes spatial memory deficits and synapse impairments with activation of GSK-3. Simultaneous inhibition of GSK-3 improves spatial memory and synaptic plasticity with mechanisms involving CREB phosphorylation at Ser129 and Ser133.

ACKNOWLEDG EMENTS
This work was supported in parts by grants from the Natural Science