Chronic diabetic states worsen Alzheimer neuropathology and cognitive deficits accompanying disruption of calcium signaling in leptin-deficient APP/PS1 mice

The coincidences between Alzheimer’s disease (AD) and type 2 diabetes mellitus (T2DM) are so compelling that it is attractive to speculate that diabetic conditions might aggravate AD pathologies by calcium dysfunction, although the understanding of the molecular mechanisms involved remains elusive. The present work was undertaken to investigate whether calcium dyshomeostasis is associated with the exacerbated Alzheimer-like cognitive dysfunction observed in diabetic conditions in APP/PS1-ob/ob mice, which were generated by crossing ob/ob mice with APP/PS1 mice. We confirmed that the diabetic condition can aggravate not only Aβ deposition but also tau phosphorylation, synaptic loss, neuronal death, and inflammation, exacerbating cognitive impairment in AD mice. More importantly, we found that the diabetic condition dramatically elevated calcium levels in APP/PS1 mice, thereby stimulating the phosphorylation of the calcium-dependent kinases. Our findings suggest that controlling over-elevation of intracellular calcium may provide novel insights for approaching AD in diabetic patients and delaying AD progression.


INTRODUCTION
Over the past two decades, intense focus has been given to investigating the relationship between Alzheimer's disease (AD) and type 2 diabetes mellitus (T2DM) [1][2][3][4][5]. According to prevalent studies, the T2DM population has a higher risk of developing AD than agematched controls [6,7], and abnormal cerebral glucose metabolism and insulin resistance have been commonly observed in both AD and T2DM [8][9][10]. Furthermore, several antidiabetic drugs have been shown to decrease the risk of cognitive decline in preclinical studies [5]. However, the mechanism for AD vulnerability in T2DM patients remains unclear.
More recently, attention has turned toward oxidative stress initiated by the formation of advanced glycation end products (AGEs) and the receptor for advanced glycation end products (RAGE), which also functions as a putative Aβ receptor, since they were identified in both AD and T2DM [11][12][13][14][15][16][17]. We have previously reported that chronic hyperglycemia not only promoted β-site cleavage of the β-amyloid (Aβ) precursor protein (APP) to increase extracellular senile plaque (SP) formation but also triggered intracellular tau hyperphosphorylation and synaptic loss in the brain, thus potentiating the cognitive dysfunction in a mouse model of combined T2DM and AD [18,19]. More importantly, we also observed that these events might be associated with the activation of RAGE signaling pathways and the elevated activation of several enzymes, such as β-site APP cleavage enzyme 1 (BACE1), presenilin 1 (PS1), glycogen synthase kinase-3β (GSK-3β), cyclin-dependent kinase 5 (CDK5), and mitogenactivated protein kinase (MAPK) [18][19][20][21]. Therefore, we hypothesize that protein glycation and increased oxidative stress via hyperglycemia are the leading causes by which T2DM increases the risk of AD [19].
In AD, the role of calcium in the brain is more clear [22]. Excessive and sustained calcium elevation can

Priority Research Paper
Oncotarget 43618 www.impactjournals.com/oncotarget alter mitochondrial oxidative phosphorylation, activate oxygenases, and then evoke free radical production. It is likely that disrupted calcium homeostasis is involved in the increased oxidative stress in neurons and other cell types, contributing to the calcium-mediated degenerative processes in AD [23][24][25]. Indeed, it has been shown that intracellular calcium overload can impair synaptic plasticity, exacerbate Aβ formation, promote tau hyperphosphorylation, trigger neuronal apoptosis, and eventually lead to deterioration of cognition via dysregulated activation of calcium-dependent kinases such as CDK-5 and Calcium/calmodulin (CaM)-dependent protein kinase II (CaMKII) [22,[26][27][28][29]. Interestingly, there is robust evidence that both in animal models of diabetes and in diabetic patients, intracellular calcium homeostasis are disturbed across both peripheral and brain tissues [30,31]. Therefore, it is reasonable to infer that dysregulation of intracellular calcium homeostasis and calcium signaling pathways represent the driving force for increased oxidative stress in T2DM contributing to the progression and development of AD.
Here, we crossed leptin-deficient mice (ob/ob, a T2DM mouse model) with APP/PS1 transgenic mice (an AD mouse model) to generate APP/PS1-ob/ob mice. Our project intended to use behavior, molecular biology, morphology and other techniques to further investigate the hypothesis that aberrant calcium signaling pathways induced by a chronic diabetic state might exacerbate AD neuropathology and cognitive deficits in APP/PS1-ob/ob mice.

Metabolic features of crossed APP/PS1-ob/ob mice
To determine the effect of diabetic symptoms on the pathogenesis of AD, we successfully generated a diabetic AD mouse model, APP/PS1-ob/ob mice, by crossing APP/ PS1 and diabetic ob/ob mice. In APP/PS1-ob/ob mice, an age-dependent excessive obesity was observed from 4 to 24 weeks of age compared with the APP/PS1 littermates (p < 0.01; Figure 1A). Importantly, APP/PS1-ob/ob mice showed severe glucose intolerance in the glucose tolerance test (GTT) when compared to the glucose intolerance of APP/PS1 mice at 12 weeks of age (p < 0.05; Figure 1C). Moreover, the blood sugar level (p < 0.01; Figure 1B) and the insulin levels of brain and serum (p < 0.05; Figure  1E and 1F) in ob/ob mice and APP/PS1-ob/ob mice were remarkably higher than those in the wild type (WT) and APP/PS1 mice, suggesting that APP/PS1-ob/ob mice also showed severe symptoms of hyperglycemia and hyperinsulinemia at 6 months of age. Interestingly, the results of the insulin tolerance test (ITT) showed that the ob/ob mice had a marked decrease in insulin sensitivity, and the APP/PS1-ob/ob mice showed a more severe sensitivity than the ob/ob mice at 12 weeks of age (p < 0.01; Figure 1D). These results indicate that the APP/PS1ob/ob mice obtained by hybridization have the typical characteristics of insulin resistance in T2DM. Therefore, the APP/PS1-ob/ob mouse is a useful animal model to study the pathophysiological relationship between T2DM and AD.
Memory deficits and cognitive disorders in APP/ PS1-ob/ob mice APP + -ob/ob mice have been reported to demonstrate early onset of AD-like cognitive dysfunction at 8 or 12 weeks [12]. To evaluate whether the diabetic condition can exacerbate the memory and cognitive damage in APP/PS1 mice, we performed the Morris water maze test (MWM) and nest construction test ( Figure 2). In the visible platform test, 6-month-old ob/ob and APP/PS1-ob/ob mice showed a longer latency escape and poorer performance (p < 0.05; Figure 2A), indicating that body weight may affect the learning ability of mice. In the hidden platform test, although the APP/PS1 and ob/ob mice showed a mild learning impairment, the APP/PS1-ob/ob mice showed a severe difference from WT mice, ob/ob mice, and even APP/PS1 mice (p < 0.05; Figure 2A and 2C). During the probe trial, the number of times that the mice crossed through the central region (where the hidden platform had previously been located) in the APP/PS1-ob/ob and APP/ PS1 groups was significantly less than that observed in the WT group. There were no significant differences between APP/PS1-ob/ob and APP/PS1 mice, although APP/PS1ob/ob mice performed worse (p > 0.05; Figure 2B). In the nest construction test, mice in the APP/PS1-ob/ob group performed worse than the APP/PS1 group mice ( Figure  2D and 2E).

Aβ generation and deposition in APP/PS1-ob/ob mice
Given the essential role of APP processing and Aβ deposition in abnormal cognitive function [25], we initially sought to elucidate the effects of a chronic diabetic state on Aβ generation and deposition in APP/PS1-ob/ob mice. Immunohistochemistry results showed a significant difference in the SP number between APP/PS1 and APP/ PS1-ob/ob mice, and the size of the Aβ-immunoreactive SP in both the cortex and hippocampus of the APP/PS1ob/ob mice was also markedly increased ( Figure 3A-3C). According to the results of increased Aβ deposition, we subsequently examined the effects of a long-term diabetic condition on APP metabolism in the APP/PS1-ob/ob mouse brain. As shown in Figure 3D, the level of total APP in APP/PS1 mice was significantly higher than that Oncotarget 43619 www.impactjournals.com/oncotarget in the WT and ob/ob mice at the age of 6 months, and the increase was markedly strengthened by an environment of continuous diabetic symptoms (p < 0.01; Figure 3D and 3E). As expected, there was a modest increase in the levels of BACE1 (the predominant β-secretase), PS1, C99 and Soluble Amyloid Precursor Protein-β (sAPPβ, the main products of amyloidogenic pathway) in the brain tissue of APP/PS1-ob/ob mice compared with those in B. Blood glucose level at 24 weeks of age. C.-D. Blood glucose levels following intraperitoneal injection of 2 g/kg glucose (C) and 0.75 U/kg insulin (D) at 12 weeks of age. E.-F. ELISA detection of insulin levels in the serum and cerebral cortex at the age of 6 months. Data represent the mean ± S.E. (n = 8). *p < 0.05 and **p < 0.01, compared with WT mice; # p < 0.05 and ## p < 0.01, compared with APP/PS1 mice; && p < 0.01 compared with ob/ob mice.
Oncotarget 43620 www.impactjournals.com/oncotarget the APP/PS1 mice, although it did not reach statistical significance (p > 0.05; Figure 3D, 3H, 3I, 3K and 3L). However, the levels of the downstream products of nonamyloidogenic α-secretase processing of APP, soluble amyloid precursor protein-α (sAPPα) and C83 (products of non-amyloidogenic α-secretase processing of APP),  Oncotarget 43622 www.impactjournals.com/oncotarget were significantly lower than those in APP/PS1 mice (p < 0.05; Figure 3D, 3F and 3G). These changes in APP processing were consistent with the decreased expression of a disintegrin and metalloproteinase domain-containing protein 10 (ADAM10, the predominant α-secretase) in the brains of the APP/PS1-ob/ob mice (p < 0.01; Figure  3D and 3J). Therefore, we can determine that the cerebral chronic diabetic condition promotes the metabolism of APP during amyloidosis, which subsequently leads to the generation and deposition of Aβ.

Deterioration of tau phosphorylation levels in the brains of APP/PS1-ob/ob mice
Tau hyperphosphorylation is one of the main pathological features of AD. To further determine whether the chronic diabetic state could aggravate the neurofibrillary tangles (NFTs) in the APP/PS1 mouse brain, we measured the tau phosphorylation level in the brain tissue of each mouse genotype. As shown in Figure 4A, leptin gene knockout induced an increase in tau phosphorylation at the location of Ser396, Ser202, Thr231, and Thr181 in these brain tissue samples; similar changes were also found in APP/PS1 mice. It is worth noting that the expression levels of p-tau (Ser396), p-tau (Ser202) and p-tau (Thr231) were significantly higher in the APP/PS1-ob/ob mice than those in the APP/PS1 mice (p < 0.05; Figure 4A-4D), whereas the increased level of p-tau (Thr181) was not significant (p > 0.05; Figure 4A, and 4E). These results are further evidence that diabetic symptoms can exacerbate tau hyperphosphorylation in the brains of mice, aggravating the development of AD pathology.

Synaptic loss in the APP/PS1-ob/ob mouse brain
Synaptic dysfunction is thought to be an early manifestation of AD. We then assessed the effect of a long-term diabetic condition on the alterations of synapses in our experimental system. As shown in Figure  5, the expression of synaptophysin (SYP) and postsynaptic density protein-95 (PSD95), which are markers of preand post-synaptic compartments, was significantly lower in APP/PS1-ob/ob mice than in APP/PS1 mice (p < 0.05; Figure 5A-5F). Notably, we also found that the chronic diabetic condition significantly reduced the expression of N-methyl-d-aspartate receptor 2B subunit (NMDAR2B), a post-synaptic marker protein regulated by calcium concentration, in both the ob/ob and APP/PS1-ob/ob mice compared with that in the WT and/or APP/PS1 mice, respectively (p < 0.05; Figure 5D and 5G), indicating that the imbalance of calcium ions may play an important role in this process. These data suggest that cognitive impairment may be closely related to synaptic loss in APP/ PS1-ob/ob mice.   Amplifying the inflammatory response in the brains of APP/PS1-ob/ob mice According to previous data, we have a hypothesis that the chronic diabetic state likely upregulates AGEs/ RAGE activity and the subsequent production of reactive oxygen species (ROS) and inflammation [12,19]. As presented in Figure 6, the ROS production in the brains of mice with diabetes was significantly higher than in the WT and APP/PS1 mice (p < 0.05; Figure 6A). Changes in the levels of AGEs and RAGE were similar to that of ROS (p < 0.05; Figure 6B-6D). As the downstream regulatory proteins of AGEs/RAGE, the activity of extracellular signal-regulated kinase (ERK) and nuclear factor-kappa B (NFκB) in the brains of APP/PS1-ob/ob mice was also obviously enhanced, which led to an increase in caspase-3 cleavage products (p < 0.01; Figure 6C, 6E-6G). These results suggest that inflammation and apoptosis may have occurred in this process. Another important source of free radicals and inflammatory response in AD comes from activated astrocytes and microglia. Western blot analyses demonstrated that, compared with that in APP/ PS1 mice, the expression levels of glial fibrillary acidic protein (GFAP) and ionized calcium-binding adaptor molecule 1 (Iba1) in the brains of APP/PS1-ob/ob mice were significantly increased (p < 0.05; Figure 6H-6J). Meanwhile, immunofluorescence results using double labeling with Aβ indicated that the activation of astrocytes and microglia cells in the brains of APP/PS1-ob/ob mice was more intensive than that in APP/PS1 mice at the age of 6 months ( Figure 6M, 6N). Next, we detected some major inflammatory cytokines. Although the expression level of interleukin 1-beta (IL-1β) and tumor necrosis factor-alpha (TNFα) was significantly higher in the brains of APP/PS1 mice than in the brains of WT mice, it is worth noting that a higher expression was found in the brains of APP/PS1-ob/ob mice ( Figure 6H, 6K and 6L).

Analysis of the signaling mechanisms of synaptic loss and tau hyperphosphorylation
In light of our finding that the chronic diabetic state increased Aβ generation and Aβ plaque formation in APP/PS1-ob/ob mice and because the expression of APP cleavage enzymes ADAM10, BACE1, and PS1 are related to Aβ synthesis and linked to the calciumsensor protein, calmodulin [32], it is plausible that the increased amyloidogenic processing and/or the inhibited non-amyloidogenic processing of APP in APP/PS1-ob/ob mice is calcium dependent. Over the past 20 years, the "calcium hypothesis" of AD has been well established [32][33][34][35], and the disturbed calcium homeostasis, which contributes to the increase in oxidative stress, may also be associated with both AD and T2DM. To determine whether the chronic diabetic state could be responsible for the pathogenesis and progression of AD, we next examined the level of calcium and changes in the specific calcium-dependent signal transduction pathway in AD neurodegeneration by examining key protein effectors in each mouse genotype.
First, we analyzed the mechanisms by which the diabetic symptoms worsen synaptic loss and tau hyperphosphorylation in APP/PS1-ob/ob mice. Several calcium-dependent protein kinases (e.g., CDK5, CaMKII, ERK, mTOR) were investigated [32,[36][37][38][39]. Given the critical roles of CDK5 in tau hyperphosphorylation, we extended our experiments to determine the signaling cascade of tau phosphorylation in APP/PS1-ob/ob mice. It is known that calpain1 can activate CDK5 by the cleavage of the CDK5 activator p35 to p25, which has a more robust activity toward CDK5 [36]. We observed a marked increase in the levels of p-CDK5 and calpain1 in the brains of APP/PS1-ob/ob mice compared with those of APP/PS1 mice, accompanied with an increased formation of p25 (p < 0.05; Figure 7A-7D). However, there were no significant differences in the levels of CDK5 and P35. Meanwhile, as presented in Figure 7, the levels of p-CaMKIIand p-ERK in the brains of APP/PS1-ob/ob mice were significantly higher than those in APP/PS1 transgenic mice (p < 0.05; Figure 7E, 7G and 7H). Interestingly, as the downstream regulatory protein of CaMKII [40], the phosphorylation level of mechanistic target of rapamycin (mTOR) was also significantly increased (p < 0.05; Figure 7E and 7F), whereas the expression levels of downstream NMDAR2B and p-CREB (c-AMP-responsive element binding protein) [41,42] in the brains of APP/PS1-ob/ob mice were significantly lower than those in APP/PS1 transgenic mice (p < 0.05; Figure 7I and 7J). Compared with those in the APP/PS1 mice, the expression levels of Brain-derived neurotrophic factor (BDNF) were not less in the APP/PS1ob/ob mice (p > 0.05; Figure 7K). Quantification showed that the levels of GFAP, Iba1 and TNFα were markedly higher in the brains of the APP/PS1-ob/ob mice than in those of the APP/PS1 mice. M.-N. Double immunofluorescence staining and confocal microscopy analyzed the distribution and expression of Aβ with GFAP and Iba1 in the cortex and hippocampus of the APP/PS1 and APP/PS1-ob/ob mouse brains. All analyses were done using 6-month-old mice. Data represent the mean ± S.E. (n = 10). *p < 0.05, **p < 0.01 compared with the WT control group; # p < 0.05, ## p < 0.01 compared with the APP/PS1 group.
Oncotarget 43626 www.impactjournals.com/oncotarget Finally, we determined the calcium concentrations in the brains of APP/PS1-ob/ob mice and WT, ob/ob, and APP/PS1 littermates at the age of 6 months. As shown in Figure 7L, atomic absorption spectrum analysis revealed that calcium levels were statistically higher in the brains of ob/ob, APP/PS1, and APP/PS1-ob/ob mice than those in the WT group. Moreover, we unexpectedly observed significantly higher calcium levels in the APP/PS1-ob/ ob mouse brain than in the APP/PS1 and ob/ob mice (p < 0.01; Figure 7L), indicating that diabetic symptoms contributed to the increase in calcium content in the brains of APP/PS1 mice.
These data revealed that calcium overload and the calcium signaling pathway may contribute enormously to the possibility of a mutual interaction between AD and DM.

DISCUSSION
Accumulating evidence supports the contribution of T2DM to AD progression. Although local inflammation, oxidative stress, impaired neurotropic factors, and AGEs have been shown to be possible common molecular mechanisms in both diseases [17,23], the underlying mechanisms by which diabetes worsens cognitive function are still unclear. Recently, altered calcium signaling has been suggested as a common proximal cause of neural dysfunction in AD [22,[43][44][45][46]. On the other hand, T2DM is also associated with neuronal calcium dyshomeostasis [30,47,48]. Along with these prior works [12,19], we developed an animal model that exhibited both diabetes and AD by crossing APP/PS1 and ob/ob mice. In the results described above, we demonstrated that a diabetic condition deteriorated cognitive dysfunction Oncotarget 43627 www.impactjournals.com/oncotarget with neuropathological changes, including Aβ and hyperphosphorylated tau accumulation, neuronal and synapse loss, and neuroinflammation in the APP/PS1ob/ob mice. Specifically, we suggest that disturbances in cellular calcium homeostasis elicit a set of biochemical cascades that may be implicated in the common pathogenesis of AD and T2DM ( Figure 8).
As reported previously [12,19], our results demonstrated that APP/PS1-ob/ob mice showed severe diabetic symptoms including hyperglycemia and insulin resistance and may be used as an animal model for researching the relationship between T2DM and AD. Moreover, we found that APP/PS1-ob/ob mice had a more severe cognitive impairment than observed in APP/PS1 mice at the age of 6 months through the action trajectory in the MWM test and behavior in the nest building experiment. As expected, we observed an increased Aβ burden in the brains of APP/PS1-ob/ob mice compared with that observed in APP/PS1 mice at the age of 6 months. Along with the increased Aβ burden, the levels of total APP were strongly augmented, and a modest increase in levels of sAPPβ and C99 was also observed. In contrast, the levels of sAPPα and C83 were markedly lower in APP/ PS1-ob/ob mice than those in APP/PS1 mice, indicating that the diabetic condition may primarily reduce the nonamyloidogenic APP cleavage pathway, thereby increasing the synthesis and accumulation of Aβ. Interestingly, the activity of ADAM10 and BACE1 is related to the calcium signaling pathway [32], and disturbed processing of APP may destabilize neuronal calcium homeostasis by increasing Aβ production and decreasing levels of sAPPα [25]. These finding may suggest that pathological changes in APP metabolism due to a diabetic condition may enhance cognitive dysfunction.
In fact, in addition to Aβ pathology, the diabetic condition could reportedly aggravate tau Oncotarget 43628 www.impactjournals.com/oncotarget hyperphosphorylation in APP/PS1 mice [19,49,50]. In the present study, we documented that tau phosphorylation at the Ser396, Ser202, Thr231, and Thr181 epitopes was dramatically increased in diabetic ob/ob mice. Notably, although AD mice exhibited obvious characteristics of tau hyperphosphorylation, the appearance of phosphorylation was more severe at the Ser396, Ser202 and Thr231 sites of tau in 6-month-old APP/PS1-ob/ob mice. Similar results have been confirmed in our previous study in the hyperglycemic Pdx1 +/− /APP/PS1 mice. Hyperphosphorylation of tau on many residues might be ascribed to the activation of multiple kinases. A previous study demonstrated that the phosphorylation of Akt, a key kinase implicated in the insulin-signaling pathway, was inhibited in 12-week-old age APP + -ob/ob mice [12], and studies that have addressed the activation of GSK-3β, one of the main physiological and pathological tau kinases, in diabetic animal models have reported contradictory findings [51]. Of note, our recent data also demonstrated that GSK-3β was not inhibited in Pdx1 +/− /APP/PS1 mice, indicating that GSK-3β is not closely linked to chronic hyperglycemia-induced tau hyperphosphorylation [19]. Thus, we examined the activation profiles of other tau kinases, including CDK5, ERK, CaMKII, and mTOR in vivo. P35 is the CDK5 activator that is cleaved by calpain1 into the P25 and P10 fragment when the calcium homeostasis of intracellular neurons is destroyed, and the CDK5-P25 complex thereby induces tau phosphorylation [52,53]. We observed that activation of CDK5 was increased following increased levels of P25 and calpain1 expression under diabetic conditions in our experiments. Moreover, we then discovered that the activation of CaMKII and mTOR, which increase synaptic damage and tau phosphorylation, respectively, was significantly enhanced in the mouse models of T2DM and AD, whereas the levels of p-CREB, the downstream protein of CaMKII, were markedly down-regulated. However, it is interesting to note that the rise in CDK5, CaMKII, and mTOR phosphorylation in APP/PS1-ob/ob mice is probably due to the disorder of calcium homeostasis, an upstream factor upregulating the phosphorylation of several kinases [22].
As an early AD pathology event, there is a strong connection between Aβ production, synaptic damage, and cognitive impairment [54]. It is likely that the perturbed processing of APP contributes to the increased production of Aβ at synapses, resulting in multiple adverse consequences on the function and integrity of both pre-and post-synaptic terminals [25]. Molecular biomarkers of synapses, such as SYP and PSD95, are good indicators of synaptic loss in AD [55][56][57]. Here, by examining the expression levels of SYP and PSD95, we found a more severe synaptic loss in APP/PS1-ob/ ob mice than that observed in APP/PS1 mice, suggesting that the diabetic condition may aggravate synaptic damage. Importantly, at postsynaptic sites, the expression of NMDARs is influenced by the interaction of PSD95 [58]. Particularly striking, NMDAR has been proposed to mediate Aβ-induced synaptic dysfunction [59,60]. In this study, NMDAR2B expression in 6-month-old ob/ ob and APP/PS1-ob/ob mice was significantly lower than that in WT and APP/PS1 mice, respectively; similar trends also occurred in the expression of p-CREB. However, although reduced BDNF levels were found in all genotypes compared to that in the WT mice, there were no significant differences in the brain BDNF levels in APP/PS1-ob/ob mice compared to that of the APP/PS1 mice. This was surprising because BDNF is a member of the neurotrophin superfamily, and recent studies suggest that lower NMDARs levels could lead to a decreased activation of the CREB pathway, subsequently blocking the production of the BDNF [42,61]. This might indicate that despite the fact that NMDAR2B levels were generally reduced in APP/PS1-ob/ob mice, some compensatory mechanisms are still in place, which might stimulate persistent NMDAR activity and induce BDNF synthesis [61], likely as an adaptive mechanism in response to the increased receptor activity due to calcium dysregulation [32,61,62].
Aβ is associated with the formation of reactive oxygen species (ROS) and is a part of a vicious cycle with other key pathological features, such as calcium dysregulation and inflammation [13,63]. The ROS results in our experiment are consistent with previous studies [19]. Increasing evidence has elucidated that both DM and AD could induce the production of ROS by the interaction of intracellular AGEs/RAGE through a nonenzymatic reaction of glucose and other carbohydrates [64][65][66][67][68]. Studies have provided data showing that AGEs can be involved in neuronal death by mediating several cell signaling cascades such as ERK and NFκB, and up regulation of RAGE and NFκB may be accompanied by overactivation of microglia, astrocytes, and proinflammatory factors [63,[68][69][70][71]. In this study, higher ERK and NFκB phosphorylation and cleaved caspase-3 levels were observed in APP/PS1-ob/ob mice than were observed in the other group, suggesting that the diabetic condition promotes autophagy and apoptosis in AD. As expected, some inflammatory factors, such as IL-1β and TNFα, were further activated in AD mice under diabetic conditions. Furthermore, in our immunofluorescence results, a greater activation of astrocytes and microglia occurred in the APP/PS1-ob/ob mice brains than in the brains of APP/PS1 mice. From these results, we can conclude that the activation of AGEs/RAGE and inflammation induced by hyperglycemia and insulin resistance may worsen the pathology of AD in this model.
One of the most important findings of the present study is that the calcium content in the brains of APP/ PS1 and ob/ob mice were significantly higher than that in WT mice, and APP/PS1-ob/ob mice exhibited an even www.impactjournals.com/oncotarget higher increase. As documented above, regardless of if the worsening pathology is due to altered APP processing, triggered tau hyperphosphorylation and synapse loss or NMDA receptor dysfunction, the potentiation of ROS production and neuroinflammation seem to be calciumdependent processes, suggesting that imbalanced calcium homeostasis may be a key factor in the aggravation of AD pathology by the diabetic condition.
In summary, growing evidence supports the contribution of altered calcium signaling to AD progression [22,72]. On the other hand, T2DM is also associated with neuronal calcium dyshomeostasis [73], suggesting that disturbances in cellular calcium homeostasis elicits a set of biochemical cascades that are implicated in the common pathogenesis of AD and T2DM (Figure 8). This study utilized the APP/PS1-ob/ob animal model with both AD and T2DM and clearly demonstrated that the diabetic condition can aggravate cognitive impairment, Aβ deposition, tau phosphorylation, synaptic loss, neuronal death and inflammation in AD mice, and damage to calcium homeostasis may promote the progress of these pathophysiological processes. Our data obtained from the APP/PS1-ob/ob hybrid mouse model clearly clarified that a chronic diabetic state can aggravate the pathology of AD. This study provides a theoretical basis for the prevention and treatment of AD.

Animals and metabolic measurements
APP/PS1 mice and ob/ob mice were initially purchased from the Jackson Laboratory (Bar Harbor, ME, USA). The former produce high levels of Aβ42 fiber deposition and express the human APPswe and PS1-A246E genes; the latter are a typical mouse model of T2DM in which the leptin gene is completely knocked out, inducing insulin resistance and high blood sugar. We subsequently intercrossed these mice to generate APP/ PS1, ob/ob, APP/PS1-ob/ob, and WT littermate mice. All mice had the same genetic background (C57BL/6). At 4 weeks of age, all animals were genotyped by polymerase chain reaction (PCR) using tail DNA. GTT and ITT were performed in 3-month-old mice. For GTT, animals received a single intraperitoneal (i.p.) injection of glucose (2 g/kg, Sigma, dissolved in normal saline) after fasting for 12 hours. For the ITT, animals fasted for 12 hours before i.p. injection of 0.75 U/kg insulin, and blood glucose levels were detected at each subsequent time point with a handheld blood glucose meter and a blood sample obtained by tail prick. At the conclusion of behavioral testing (at 6 months of age), all genotypes of mice were sacrificed under sodium pentobarbital (50 mg/kg, i.p.) anesthesia. Just before decapitation, a whole blood sample was collected directly from the heart, and the serum was separated from the whole blood. Separated serum was stored at -80 o C for subsequent ELISA detection. Then the brains were bisected sagittally after perfusion with 0.9% normal saline. One hemisphere was stored at -80 °C for extracting protein, and the other hemisphere was immersed in 4% paraformaldehyde (for 48 hours) for frozen and paraffin sections. All experimental procedures were approved by the Laboratory Animal Ethical Committee of Northeastern University.

Morris water maze
The week before the mice were sacrificed, 23-weekold mice were trained and tested in a Morris water maze (MWM), as detailed in our previous studies [74]. Subsequently, the activity of the mice was recorded and analyzed with a computer program (ZH0065; Zhenghua Bioequipment).

Nest construction
To assess the social behavior of animals, the nest construction test was carried out as detailed in our previous study [75]. The nests were scored the following morning according to a 4-point system: 1, no biting/ tearing, with random dispersion of the paper; 2, no biting/ tearing of paper, with gathering in a corner/side of the cage; 3, moderate biting/tearing of paper, with gathering in a corner/side of the cage; and 4, extensive biting/tearing of paper, with gathering in a corner/side of the cage.

Immunohistochemistry and immunofluorescence
Paraffin-embedded brain tissue samples were sectioned (5 μm) and dewaxed, followed by antigen retrieval using L.A.B solution (Polyscience, Inc) for 15 minutes. Soon afterwards, the tissue samples were incubated with mouse anti-Aβ (1:200; Santa Cruz Biotechnology) or rabbit anti-synaptophysin (SYP, 1:200; Sigma) at 4 °C overnight. After the tissue samples were rinsed with PB (0.01 mmol/L) and incubated in secondary antibodies, they were immunostained in 0.025% DAB for immunohistochemical detection. Slides were sealed with resin to be viewed under the microscope. For double immunofluorescence, frozen sections (10 μm) were incubated with mouse anti-Aβ (1: 200; Santa Cruz Biotechnology) and rabbit antiglial fibrillary acidic protein (GFAP, 1:100; Sigma) or mouse anti-Aβ (1:200; Santa Cruz Biotechnology) and rabbit anti-ionized calcium-binding adaptor molecule 1 (Iba1, 1:100; Sigma) primary antibodies overnight at 4 °C. Secondary antibodies, donkey anti-mouse IgG conjugated with fluorescein isothiocyanate (1:200; www.impactjournals.com/oncotarget was performed using a Bio-Rad system, and grayscale analysis was performed using Image J software. The intensity of each band is divided by its own β-actin, and then the relative value was normalized to that of WT group.

Analysis of calcium content
Brain tissue samples were accurately weighed for the determination of calcium content in each genotype of mice (n = 10). Samples were treated with 100 µl nitric acid (sigma, Purity ≥ 90%) at 100 °C for 15 minutes. After cooling to room temperature, the samples were fixed to a 10-ml volume with 1% nitric acid. A flame atomic absorption spectrometer (ZEEnit700P, Analytikjena, Germany) equipped with monoelement hollow cathode lamps was used, and a high purity acetylene gas was used as fuel. The determination of the total calcium content was carried out under the following conditions: acetylene-air flow rate 65 L/h, burner height 5 mm, slit width 0.8 mm and wave length 422.7 nm. All items were detected under the optimal conditions, and 5 replicates were used.

Measurement of insulin and AGEs levels
For quantification of insulin levels, mouse serum was obtained by placing the blood at room temperature for 20 minutes. The supernatant was then obtained after 4000 rpm/min centrifugation of the blood for 10 minutes. Brain tissue was obtained as described before. The amount of AGEs and insulin was detected by enzymelinked immunosorbent assay (ELISA) kits (LanpaiBIO, Shanghai, China) in a fasted state, according to the manufacturer's instructions. The absorbance was measured using a BIO-RAD 3550-UV microplate reader.

Statistical analyses
All values are presented as the mean ± standard error of the mean (SEM). Repeated measures analysis of variance (ANOVA) was performed for the MWM tests; differences among the means were evaluated with multivariable ANOVA. Other comparisons were analyzed by two-way ANOVA followed by post hoc Bonferroni tests when appropriate. All data were analyzed using SPSS 16.0 software, and differences were assumed to be highly statistically significant if p < 0.01 and statistically significant if p < 0.05.

Author contributions
Shuai Zhang, mainly researched data; Rui Chai, researched data; Ying-Ying Yang, researched data; Shi-Qi Guo, researched data; Shan Wang., Tian Guo., and Shuang-Feng Xu, generated and validated the mouse model; Yan-Hui Zhang, contributed to discussion; Zhan-You Wang, reviewed/edited manuscript, Chuang Guo, designed/wrote manuscript. Dr. Chuang Guo is the guarantor of this work and, as such, had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis.