Effects of Distinct n-6 to n-3 Polyunsaturated Fatty Acid Rations on Insulin Resistant and AD-like Phenotypes in High-Fat Diets-Fed APP/PS1 Transgenic Mice

Xiaojun Ma Capital Medical University Yujie Guo Capital Medical University Pengfei Li Capital Medical University Jingjing Xu Capital University of Medical Sciences: Capital Medical University Shengqi Dong Capital Medical University Yanyan Gao Capital Medical University Nicholas Van Halm Lutterodt Capital Medical University Linhong Yuan (  ylhmedu@126.com ) Capital Medical University https://orcid.org/0000-0001-5169-2527

etiology of AD, as well as revealing the fundamental pathophysiological mechanisms of insulin resistant related dementia.

Animals and diet
A total of 36 male APP swe /PS1 dE9 transgenic mice and 6 sex-matched C57BL/6J control mice (all at 4 weeks old) were purchased from the experimental animal center of Capital Medical University. The mice were maintained on a 12:12 light-dark cycle and allowed access to water and food ad libitum. Experiment protocols were licensed according to Capital Medical University Animal Care and Use Committee regulations and associated guidelines (AEEI-2019-071). All diets were purchased from SYSE Bio-tech. Co.
LTD, Changzhou, China. The normal diet (ND) and HFD differ in the composition of fat added to the diet. Full details of the diet are provided in Table 1.

Treatment
According to baseline fasting blood glucose level, 36 male APP/PS1 mice were randomly split into 2 treatment groups: (1) ND group (n = 6): the mice were fed with a standard diet, in which 10% kcal are derived from fat chow (10 kcal%, Product code: D18091704); (2) HFD group (n = 30): in this group, the mice were fed with high-fat diet, in which 60% kcal are derived from fat (60 kcal%, Product code: D12492). Sucrose content per gram was equivalent between the two diets. Diet intakes of animals were monitored every 3 days, and body weights were assessed weekly. Plasma glucose levels of APP/PS1 mice treated with 60% HFD were monitored monthly. After 60% HFD dietary intervention for 3.5 months, glucose tolerance test (GTT) was performed to assess insulin resistance phenotype. Any mouse that did not show an increase in plasma glucose in comparison with control mice was excluded from the study. At the age of 4.5 months, the mice with increased fasting glucose level were further randomly divided into 4 subgroups (at least 6 mice for each group): (1) the high n-6/n-3 PUFA ratio diet group: the mice were provided with 45% HFD with n-6/n-3 = 16:1 (Product code: D18091701); (2) the medium n-6/n-3 PUFA ratio diet group: the mice were treated with 45% HFD with n-6/n-3 = 5:1 (Product code: D8091703); (3) the low n-6/n-3 PUFA ratio diet group: the mice were treated with 45% HFD with n-6/n-3 = 1:1 (Product code: D18091702); (4) the 60% HFD group: the mice were fed with their original 60% HFD diet. The control APP/PS1 mice and C57BL/6J mice were maintained on their original control diets. After 3.5 months under these dietary regimens, the mice were used for GTT and behavioral measurement. The whole schematic diagram of animal dietary intervention is shown in Supplementary Fig. 1.

Fasting blood glucose levels and GTT
Blood glucose levels were measured per month, using a standard glucometer (Accu-Chek, Roche) with blood obtained from the tail tip. GTT were performed at the time point of 60% HFD intervention for 14 weeks, and the end of the experiment.

Behavioral testing
At the end of the dietary intervention, the spatial learning and memory abilities of animals were analyzed by Morris water maze as described previously [16]. The escape latency, path length travelled to arrive at the platform, average swim speed, and platform site crossings were recorded with a video tracking system (Water Maze 2.6 Institute of Materia, Chinese Academy of Medical Sciences DMS-2, Beijing, China).

Tissue preparation
Following behavioral testing, all mice were euthanized and sacri ced. Their brains were removed and separated along the midsagittal sulcus. Half of each brain was immediately placed in a cold saline solution and then the cortical and hippocampal regions were dissected and stored at -80℃ until used for biochemical measurements. The other half side of the brain was used for the histological studies. Other tissue samples including liver, muscle and adipose tissue were frozen at -80℃ for biochemical assays or xed in formalin for histological analyses.
Histology, immuno uorescent and immunohistochemical assay The liver samples were embedded in 10% formaldehyde buffered solution, subsequently embedded in para n blocks, and sectioned into slices of 5 µm, and then stained with hematoxylin-eosin (HE). Specimens were examined under a light microscope. Beta-amyloid plaque in the cortex and hippocampus was measured by using Congo red (KeyGEN BioTECH, Nanjing, China) staining kit.
Immunohistochemistry (IHC) assay was applied for measuring glucose transporters (GLUTs) in tissues. Brie y, brain, liver, skeletal muscle and omental fat tissue samples were xed for 48 hours, embedded in para n, and then sectioned into slices of 5 µm. IHC was performed on para n-embedded sections using antibodies directed against GLUT2, (1:100 dilution, Cell Signaling Technology, USA), GLUT3, (1:100 dilution, Cell Signaling Technology, USA), and GLUT4 (1:100 dilution, Proteintech, USA). Secondary DAB antibody was used for positive detection. Then, the section was observed and photographed by electric microscope (OLYMPUS BX61).

Cortical fatty acids pro le measurement
The fatty acid concentration in the cortexes was determined according to the method described previously [17]. Fatty acid compositions of 3-4 brains per group were reported, and the fatty acids contents were expressed as g/100 g total fatty acids.

ELISA and biochemical assay
Serum TNF-α and IL-6 contents were measured using commercial ELISA kits (Immunoway, Suzhou, Western blotting A total of 20 µg of protein lysis extracts were electrophoresed on Tris-tricine gradient gels as described previously [18]. And primary antibodies used in western blotting were shown in Supplementary Table 1. RNA extraction and qRT-PCR Cortical total RNA was extracted using Ultrapure RNA Kit (Cwbio, Jiangsu, China) according to the manufacturer's protocol. Reverse transcription and target gene ampli cation assays were performed as described previously [18]. The primers used for PCR were shown in Supplementary Table 2.

Statistical analysis
The results were reported as mean ± standard deviation (SD), and P < 0.05 was considered statistically signi cant. The data were calculated and analyzed by Prism (GraphPad Software Inc., San Diego, CA, USA). Comparisons between controls and treated groups were determined using one-way ANOVA followed by Bonferroni test.

Bodyweight
Changes in body weight for the animals from different groups are depicted in Fig. 1A. During the rst 3.5 months, 60% HFD-fed APP/PS1 mice showed signi cantly higher body weight than ND-fed C57BL/6J and APP/PS1 control mice. When changing the diet to 45% HFD, we found that the bodyweights of APP/PS1 mice decreased dramatically in comparison with 60% HFD-fed APP/PS1 mice, especially in mice treated with a low n-6/n-3 PUFA ratio diet; however, the bodyweights of these mice were continuously higher than ND-fed animals.
Serum glucose, insulin levels and lipid pro le Following 60% of HFD intervention for 3.5 months, the mice showed a signi cant increase of fasting blood glucose level and area of under glycemic curve (AUC) than normal diet-fed C57BL/6J and APP/PS1 mice (P < 0.05) (Fig. 1B). After changing 60% HFD to 45% HFD, a decrease in AUC value was observed in the APP/PS1 mice compared with the 60% HFD-fed animals, especially in the 45% HFD and 45% HFD-N3 diet-fed mice (P < 0.05) (Fig. 1D). Moreover, at the end of the experiment, 60% HFD-fed animals showed a signi cant increase in serum insulin level in comparison with 45% HFD-fed animals, and statistical signi cance was also observed between 60% HFD, 45% HFD and 45% HFD-N6 groups (P < 0.05) (Fig. 1E).
The serum lipid pro le was presented in Table 2. We found that 60% HFD treatment signi cantly increased serum TC, LDL-c and high-density lipoprotein cholesterol (HDL-c) levels in APP/PS1 mice (P < 0.05). After changing the 60% HFD diet to 45% HFD, we observed a signi cant decrease in serum TC and HDL-c levels, and the mice fed with 45% HFD-N3 diet showed the lowest serum TC level (P < 0.05). Decrease in serum LDL-c level was only observed in 45% HFD-N3 diet-fed APP/PS1 mice (P < 0.05). C57BL/6J ND: C57BL/6J control mice treated with normal control diet; APP/PS1 ND: APP/PS1 mice treated with normal control diet; APP/PS1 60% HFD: APP/PS1 mice treated with 60% high fat diet for 7 months; APP/PS1 HFD-N3: APP/PS1 mice treated with mice 60% high fat diet for 3.5 months, then 45% high fat diet with n-3 PUFAs was supplied for another 3.5 months; APP/PS1 45% HFD: APP/PS1 mice were treated with 60% high fat diet for 3.5 months, then 45% high fat diet were provided for another 3.5 months; APP/PS1 45% HFD-N6: APP/PS1 mice were treated with 60% high fat diet for 3.5 months, then 45% high fat diet with n-6 PUFAs supplement were provided for another 3.5 months.

Hepatic histology
Compared with C57BL/6J control mice, the APP/PS1 control mice showed relatively higher lipid deposition in the livers. At the end of dietary intervention, the mice in the 60% HFD group showed a signi cant increase in hepatic lipid deposition. In contrast, hepatic lipid deposition was signi cantly lessened in mice treated with 45% HFD animals, especially in the n-3 PUFAs rich diet-treated mice (Fig. 1F).

Behavior
As showed in Fig. 2A, after 4 days of training, both control mice and HFD-fed APP/PS1 mice showed a signi cant decrease in escape latency. On day 5, the mice treated with n-3 PUFA rich 45% HFD showed increased escape latency than mice from other groups, however, we did not observe any statistical signi cance between the groups (P > 0.05). Additionally, the APP/PS1 mice treated with 60% HFD and n-3 PUFA rich diet showed relatively shorter platform crossing times on day 5 than mice from other groups, but the differences between groups are not statistically signi cant (P > 0.05).

AD pathology
As shown in Fig. 2B, in comparison with C57BL/6J control mice, typical Aβ plaque deposition was found in the cortexes and hippocampi of APP/PS1 control mice. The treatment of 60% HFD seemed to have no signi cant effect on the deposition of Aβ plaque in cortexes and hippocampi of the APP/PS1 mice. Diets with different n-6/n-3 PUFA ratios also had no signi cant impact on cortical and hippocampal Aβ plaque deposition in the APP/PS1 mice.
Cortical LDL-c and TC levels are shown in Fig. 2D. Compared with the control mice, the mice fed with 60% HFD showed a signi cant increase of cortical LDL-c content (P < 0.05). In contrast, 45% HFD-fed mice showed lower cortical LDL-c levels as compared with mice from 60% HFD-fed group, and the n-6 PUFAs rich 45% HFD showed the lowest cortical LDL-c content than mice in other HFD-fed groups (P < 0.05). HFD treatment did not affect cortical TC levels, and no statistical signi cance was observed between the groups (P > 0.05).
Cortical amyloid precursor protein (APP) and β-secretase-1 (BACE1) protein expression were shown in Fig. 2E. APP/PS1 control mice showed higher cortical APP protein expression than C57BL/6J control mice. The treatment of 60% HFD further up-regulated the cortical APP protein expression in APP/PS1 mice. After changing 60% HFD diet to 45% HFD diet, the decreased APP protein expression was observed in the cerebral cortex of APP/PS1 mice, especially in the mice from 45% HFD and 45% HFD-N6 groups. The dietary intervention did not affect cortical BACE1 protein expression, and no difference was observed between the groups (P > 0.05).
GLUT3 and insulin metabolism-related molecular in the cortex Cortical GLUT3 protein expression was shown in Fig. 3A. APP/PS1 control mice displayed lower cortical GLUT3 protein expression in C57BL/6J control mice. The treatment of 60% HFD signi cantly increased the cortical GLUT3 protein expression in APP/PS1 mice. After replacing 60% HFD with 45% HFD, cortical GLUT3 protein expression was down-regulated and the mice from n-3 PUFAs rich 45%HFD group showed relatively lower cortical GLUT3 expression than mice treated with 45% HFD and n-6 PUFAs rich 45% HFD.
Cortical insulin level was shown in Fig. 3B. The APP/PS1 control mice showed similar cortical insulin level as the C57BL/6J control mice and the treatment of HFDs did not affect cortical insulin level in APP/PS1 mice, although the mice from the 45% HFD group showed decreases in cortical insulin levels, no statistical signi cance was observed between the groups (P > 0.05).
High-fat diet treatment did not affect cortical insulin receptor (IR), insulin receptor substrate 1 (IRS-1), insulin degrading enzyme (IDE) and phosphorylated-insulin receptors (p-IR) protein expressions in APP/PS1 mice. The n-6 and n-3 rich 45% HFD-fed mice showed a signi cant decrease of cortical p-IR/IR ratio, and the mice fed with n-6 PUFA rich 45% HFD showed the lowest cortical p-IR/IR ratio than mice from the other groups (P < 0.05) (Fig. 3C & 3D).
As shown in Fig. 3E, APP/PS1 control mice showed relatively higher cortical glycogen synthase kinase-3β (GSK-3β), but lower p-GSK3β/GSK-3β ratio than C57BL/6J control mice. The 60% HFD treatment caused a decrease of cortical phosphorylated-glycogen synthase kinase-3β (p-GSK3β) protein expression in APP/PS1 mice, but no effect on cortical GSK3β protein expression. The n-6 and n-3 rich 45% HFD caused a signi cant decrease of cortical GSK3β protein expression, and the lowest p-GSK3β protein expression was observed in n-6 rich 45% HFD-fed mice (P < 0.05). The ratio of p-GSK3β to GSK3β showed a decreasing trend in HFD-treated animals, and the mice from the 45% HFD-N6 group exhibited the lowest cortical p-GSK3β/GSK-3β ratio (P < 0.05).
Expression of cortical NF-κB p65 and the mRNA of in ammatory cytokines The expression of cortical NF-κB p65 protein expression was shown in Fig. 4A. The red uorescence was observed in the cortices of all mice groups, indicating that NF-κB was abundantly expressed in the cortex.
Meanwhile, mice from 60% HFD-fed group and n-6 PUFAs rich 45% HFD-fed group exhibited higher NF-κB p65 expression in the nuclei than mice in other groups, indicating a migration of NF-κB p65 from the cytoplasm to nuclei.

Discussion
In the current study, AD model mice (APP/PS1) were treated with a HFD to induce insulin resistance and the effect of diet with different n-6/n-3 PUFAs rations on insulin resistance phenotype and AD-like pathology was explored. Our data showed that HFD e ciently induced insulin resistance in APP/PS1 mice, which was indicated by the dramatic increase in body weight, blood glucose, lipids and insulin levels, glucose tolerance, and severe hepatic lipidosis as compared with the normal diet-fed control mice.
We also found that a reduction of fat-derived energy (as demonstrated with a 45% HFD) was e cient in reversing the insulin resistant phenotype that was initially caused by the 60% HFD in APP/PS1 mice.
Dietary n-6/n-3 PUFA ratio-dependent regulatory effect of HFD on insulin resistance has been previously described, in which the author reported that rats fed with HFD containing n-3 PUFAs (n-6/n-3 = 1:1) showed signi cantly higher insulin sensitivity than rats treated with high n-6 PUFA HFD (n-6/n-3 = 4:1) [19], suggesting that an increase in dietary n-3 PUFAs proportion could improve insulin sensitivity and prevent HFD-induced insulin resistance. It has also been postulated that the visceral adipose tissue (e.g., fatty liver) induced increased circulating free fatty acids (FFAs) and triglycerides, released a signi cant amount of pro-in ammatory cytokines, and these cytokines disrupt the insulin action, consequently leading to decreased insulin sensitivity and resistance [20]. Consistent with these studies, our data indicated that diets with different n-6/n-3 PUFAs rations discrepantly affected insulin resistance phenotype in AD model mice, which was demonstrated by the signi cant decrease in body weight, serum TC, LDL-c levels and hepatic lipid deposition in the initial HFD-induced APP/PS1 mice later treated with high n-3 PUFA diet (n-6/n-3 = 1:1). These results align well with the previous report, which showed that n-6 PUFA rich diet-fed animals were more likely to manifest liver steatosis [21], while n-3 PUFAs intervention signi cantly reduced serum TC and LDL-c levels [22], and signi cantly relieved the signs of HFD-mediated hepatic steatosis. Unexpectedly, in our study, we did not observe the different re ection of serum glucose and insulin levels to diets with different n-6/n-3 PUFA ratios, demonstrating that the regulation of lipids metabolism may partly account for the discrepant insulin resistance relieving effect of the high n-3 PUFA diet (n-6/n-3 = 1:1) in AD model mice.
A study has shown that an HFD resulted in a signi cant increase in plasma IL-6 and TNF-α levels in APP/PS1 mice [23]. Consistent with this report, we observed signi cantly increased levels of serum IL-6 and TNF-α in 60% HFD-fed mice, and the elevation of serum pro-in ammatory markers were dramatically lowered following treatment with 45% HFD. Moreover, dietary n-6/n-3 ratio-dependent inhibition of serum TNF-α level was also observed, which was indicated by the lowest serum TNF-α level in n-3 PUFA rich dietfed APP/PS1 mice. High dietary n-6/n-3 PUFAs ratio has been reported to correlate with elevated plasma concentration of IL-6 and TNF-α in rats [24]. High n-6/n-3 PUFAs ratio (4:1) HFD-fed SD rats also displayed higher serum IL-6 and TNF-α than low n-6/n-3 ratio (1:1) HFD-fed animals [19]. Reduced IL-6 and TNF-α gene expression have been reported in skeletal muscle tissue of diabetic rat-supplemented with n-3 PUFA [25]. All these results suggested that n-3 PUFA enriched diet might antagonize HFD-induced pro-in ammatory response, and dietary n-6/n-3 PUFAs ratio might have potent modulatory effects on in ammatory reaction in experimental animals.
Reduced n-3 PUFA levels have been found in the brain (mainly in the hippocampus) of AD patients, and the increase of n-6/n-3 PUFAs ratio in the brain has been considered to be associated with the increased risk of AD [26]. In the current study, we did not observe a signi cant difference in cortical PUFA between C57BL/6J control mice and APP/PS1 control mice, indicating that the mutation of APP gene did not alter the cortical PUFAs pro le in the AD model mice. Of note, the treatment of HFD caused an increased trend in cortical PUFAs pro le in APP/PS1 mice, although the difference was not statistically signi cant when compared with control mice. Importantly, our data indicated the potential effect of n-6 PUFA rich diet on elevating cortical n-6 PUFA level and n-6/n-3 PUFAs ratio in APP/PS1 mice, Moreover, this increase of cortical n-6 PUFA level and n-6/n-3 PUFAs ratio effect could be antagonized by 45% HFD (n-6/n-3 = 5:1) and 45% HFD-N3 (n-6/n-3 = 1:1) diets. All these results indicated a strong correlation between cerebral PUFAs pro le and dietary fatty acids status, indicating a potential modifying role of dietary n-6/n-3 PUFAs ratio on cerebral n-6 and n-3 PUFA levels.
The senile plaque remains the pathological hallmark sign in AD process, and results from clinical studies have also shown that n-3 PUFA interventions can alleviate AD pathological changes and prevent synaptic degeneration [27]. Data from experimental animals also found that n-3 PUFAs rich diet has been effective in inhibiting the formation of amyloid plaque in the hippocampus and cortex in APP transgenic mice [28]. In our study, although the number of plaques in 45% HFD-N3 diet-fed APP/PS1 mice showed a slightly lesser number of senile plaques than mice treated with control diet, 60% HFD and 45% HFD-N6 diet, no statistical signi cance was observed between groups. We further detected cortical soluble and insoluble Aβ 1−40 and Aβ 1−42 levels. Consistent with a previous report [29], we found that HFD treatment increased cortical soluble Aβ 1−40 and Aβ 1−42 levels in APP/PS1 mice, indicating that HFD might promote the generation of neurotoxic soluble Aβ 1−42 and potentially accelerate synaptic degeneration and AD pathological process. It was reported that the excessive increase of lipids (especially cholesterol) in the brain could modify the composition and function of lipid raft in neuronal membranes and enhance βsecretase-mediated APP processing, ultimately promoting over-generation of Aβ [30]. Thus, we further examined cortical cholesterol levels in mice treated with different diets. The dramatic increase in cortical LDL-c was observed in 60% HFD-fed APP/PS1 mice, but HFD forti ed with PUFAs signi cantly reduced cortical LDL-c levels, especially in HFD-N6 diet-fed group. This nding is consistent with a previous study, that identi ed n-6 PUFA as a potent down-regulator of LDL-c level and alleviating depressed LDL receptor activity while promoting the diminishment of circulating LDL-c [31]. We, therefore, speculated that the disturbance of cerebral lipid hemostasis in HFD-fed animals might contribute to the increased cortical soluble Aβ content observed in the AD model mice. Ettcheto et al. demonstrated that HFD-fed APP/PS1 mice showed signi cantly higher cerebral insoluble Aβ 1−42 content than ND-fed control animals [32]. On the contrary, we found that 60% HFD treatment decreased cortical insoluble Aβ 1−40 and Aβ 1−42 levels. A study conducted in FAT-1 mice has found that n-3 PUFA enriched diet could promote the clearance of Aβ from the brain through the cerebral lymphatic system [33]. In our study, we observed an n-6/n-3 ratiodependent effect of HFD on regulating cortical soluble and insoluble Aβ levels, which was demonstrated by lower soluble and insoluble Aβ levels in mice treated with 45% HFD-N3 and 45% HFD-N6 diets. Also, consistent with the change in LDL-c level as documented earlier, we found that the diet with the highest dietary n-6/n-3 ratio caused the most signi cant decrease in soluble and insoluble Aβ levels in the cortex, suggesting the important role of n-6/n-3 PUFA ratio in modulating cerebral Aβ 1−40 and Aβ 1−42 levels. Our ndings also imply that speci c PUFAs (n-3 or n-6) may be critically required by cerebral components to execute optimal brain function(s) and that could account for the observed discrepant levels of cortical Aβ and LDL-c in response to HFD forti ed with high n-6/n-3 PUFAs ratio. Further studies are needed to uncover the underlying mechanisms behind the observed discrepant response of Aβ levels to dietary n-6/n-3 PUFA ratio.
It is well known that BACE1 catalyzes the formation of the amyloid-beta peptide from the APP. An increase in BACE1 activity enhanced the production of Aβ and triggered the deposition of Aβ plaque in the brain [34]. In our study, we only observed a decreased trend of APP in mice treated with different n-6/n-3 PUFA ratios without affecting the protein expression of BACE1. This nding suggests that the APP was modulated by PUFAs in HFD, and this modulatory effect was exclusive to BACE1 gene activity. The mismatched changes in APP and BACE1 protein expressions and cerebral Aβ content suggest the involvement of other potential molecular mechanisms in the homeostasis of Aβ in the brain.
The IR is a kind of receptor from the tyrosine kinase family. After binding with insulin, IR activates intracellular tyrosine kinase and initiates intracellular signaling pathways involving a series of physiological responses. Reduced IR sensitivity and hyperphosphorylation of IR have been reported in AD patients [35], potentially suggesting that impaired insulin responsiveness in the brain may be intrinsic to AD pathogenesis. In the current study, we found that the mice treated with n-3 and n-6 PUFA rich diets showed a signi cant decrease in cortical p-IR/IR ratio. The decrease of p-IR/IR ratio indicates the inhibition of insulin metabolic pathway, which in turn inhibits the glucose uptake. Vishal Kothari et al. also reported that the HFD-fed C57BL/6J mice exhibited a lower p-IR/IR ratio than the normal diet-fed control animals [36]. Contrary to that nding, in our study, 60% and 45% HFD did not cause a signi cant decrease of cortical p-IR/IR ratio, however, a decrease of cortical p-IR/IR ratio was observed, especially in mice treated with n-6 PUFA rich 45% HFD, suggesting that dietary PUFAs have the potential in regulating phosphorylation of IR in HFD-fed APP/PS1 mice. It has been reported that increasing pro-in ammatory cytokines (such as TNF-α and IL-6) can potentially disorient insulin metabolic action(s) and receptor functioning [37]. Excessively high n-6 PUFAs was reported to accelerate oxidative stress and oxidized LDL-c [38]. All, these results could partially explain the decrease of cortical p-IR/IR ratio which was especially observed in mice treated with n-6 PUFA rich 45% HFD.
Insulin mRNA expression has been found in both human and mouse brain samples. Rhea et al. also reported that insulin can transport across the blood-brain barrier (BBB) in vivo with either loss or inhibition of the signaling-related insulin receptor, but the binding of insulin to the brain endothelial cells would be decreased [39]. In our study, APP/PS1 mice showed similar cortical insulin levels with C57BL/6J control mice, indicating that the AD pathological changes did not affect the insulin content in the AD model mice.
Also, we did not observe any alteration in cortical insulin level in mice treated with HFD. Stanley et al. reported that peripheral hyperinsulinemia could not affect cerebral insulin signaling and insulin levels [40]. Moreover, a study has proved that cells from the hippocampus and olfactory bulb can produce insulin [41]. Our study further con rms that the cerebral insulin homeostasis cannot be affected by peripheral hyperinsulinemia in APP/PS1 mice, and we reasonably hypothesized that this restriction to the hyper-permeability of insulin to the cortex may be tightly regulated by the BBB. The inconsistent changes of serum and cortical insulin level in response to HFD treatment indicated a discrepant regulating mechanism of peripheral and central insulin metabolism. Further investigations are needed to elucidate the underlying mechanism attributing to the discrepant peripheral and central insulin metabolism.
Glucose uptake in the brain is dependent on IR-stimulated translocation of glucose transporter in the plasma membrane. GLUT3 is extensively expressed in the cortex and plays an important role in neuronal glucose uptake. Evidence suggests that HFD has the potential of reducing cerebral GLUT3 protein expression and causing impaired cerebral glucose uptake [36]. Conversely, Kothari and colleagues' study reported a compensatory overexpression of GLUT3 in the brains of diabetic rats [42]. In the present study, APP/PS1 mice showed lower cortical expression of GLUT3 protein than C57BL/6J control mice, which indicated an AD-like pathology-mediated decrease of cerebral glucose uptake in the AD-model mouse. We also found that 60% HFD-fed APP/PS1 mice displayed a signi cant increase of cortical GLUT3 expression, suggesting that HFD was e cient in reversing the downregulated glucose uptake in the brain in AD model mice. These results indicated that the increased cortical expression of GLUT3 protein expression could be a compensatory genetic positive-feed response to the insulin resistance initially induced by the HFD, and potentially due to a decreased threshold to cortical insulin sensitivity (since the primary energy source to the brain is obtained from glucose). Prolonged exposure to HFD not only induces pro-in ammatory cytokines and insulin resistance but may also be involved in adaptive genetic feedback response to engage essential proteins and transporters to compensate and improve glucose or nutrient uptake into cells as we have observed with GLUT3 in these HFD-induced insulin resistant APP/PS1 mice. Relative lower cortical GLUT3 expression was found in APP/PS1 mice from 45% HFD-N3 group as compared with mice treated with 45% HFD and 45% HFD-N6 diets, indicating a dietary PUFAsdependent optimal response of cerebral GLUT3 expression. On the contrary, we detected higher GLUT3 expression in the liver and GLUT4 expression in muscle in APP/PS1 control mice in comparison with C57BL/6J mice ( Supplementary Fig. 2A&B). This observation suggests that there may be an overlap of peripheral (hepatic and skeletal) insulin resistance phenotype and AD-like pathological cascade in these ND-fed APP/PS1 mice, implying that AD pathologic process (particularly, APP gene dysfunction) may be a risk factor for peripheral insulin resistance. Further study is required to ascertain this observation. We, however, observed a signi cant decrease of these GLUTs in 60% HFD-fed APP/PS1 mice in liver and muscle tissues (potentially due to a slightly increased threshold to peripheral insulin sensitivity) further indicating a discrepant peripheral and central glucose metabolism in response to HFD intervention in APP/PS1 mice. Additionally, the divergent sensitivities of liver or muscle GLUT3/4 expression in mice treated with the different diets also hint the role of dietary PUFA rations on glucose uptake in peripheral tissues.
IRSs are known to be major substrates of receptor kinases, mediating insulin-like growth factors (IGFs)/insulin signals to direct anabolic bioactivities, such as energy and glucose homeostasis, growth and adipocyte differentiation [43]. Postmortem studies have described a reduction in brain expression of IR adapter protein IRS-1 in AD subjects [44]. In comparison with C57BL/6J mice, signi cantly upregulated IRS-1 protein expression was found in the hippocampus of APP/PS1 mice [45]. The IDE is responsible for insulin breakdown in various tissues. Results from other studies have indicated that HFD-fed APP/PS1 mice showed signi cantly lower brain IDE protein expression than normal diet-fed control animals [46].
We found that the IRS-1 and IDE cortical protein expression remained unchanged in APP/PS1 mice treated with different diets. These results indicate that the effect of n-6/n-3 PUFA ratios in HFD on cerebral glucose homeostasis and insulin signaling pathway in APP/PS1 mice are not entirely attributable to the regulation of IRS-1 and IDE expression. The other mechanisms, such as the regulation of enzyme activity, interactions or binding between substrates and enzymes, might contribute to the HFDmediated changes of central insulin metabolism, warranting further exploration.
The GSK3β is a key molecule involving in neural progenitor cell proliferation, neuronal polarity, and neuroplasticity [47]. Insulin could stimulate phosphorylation of GSK3β and consequently result in bidirectional (phosphorylation and inactivation) effect on glycogen synthase, an enzyme responsible for glycogenesis. Elevated total GSK3β protein expression indicates hyper-activity of GSK3β signaling [48]. In the pathology of AD, GSK3β has been shown to promote Tau phosphorylation through the insulin/phosphoinositide 3-kinase (PI3K)/protein kinase B (Akt) signaling pathway [49]. In our study, APP/PS1 mice showed a relatively higher cortical GSK3β expression but lower p-GSK3β/GSK3β ratio than C57BL/6J control mice, indicating a potential activation of the GSK3β signaling pathway in this AD model mice. HFD treatment has been reported to cause a decrease in cortical p-GSK3β expression and p-GSK3β/GSK3β ratio in APP/PS1 mice, suggesting a regulatory effect of HFD on the phosphorylation of GSK3β and in modifying the activity of GSK3β. Moreover, a decreased trend of cortical GSK3β expression was found in 45% HFD-N3 and 45% HFD-N6 diets treated mice. Additionally, signi cant decreases of cortical p-GSK3β and p-GSK3β/GSK3β ratio were also found in 45% HFD-N6 diet-fed APP/PS1 mice, indicating hyper-activation of GSK3β in these high n-6/n-3 PUFA ratio diet-fed APP/PS1 mice. Together with the relatively high cortical GLUT3 protein expression observed in these animals, we speculate that diet with a high n-6/n-3 PUFA ratio might affect brain glucose metabolism through regulating the phosphorylation of GSK3β, therefore affecting the uptake of glucose in the brain. Altogether, our ndings indicated that the discrepant regulation of GSK3β activation might contribute to the dietary n-6/n-3 PUFA ratio-associated modifying effect in insulin signaling pathway in the brain of APP/PS1 mice.
NF-κB is a nuclear transcription factor that belongs to the Rel protein family, and accumulating evidence suggests its critical role in the pathophysiology of AD. Activated NF-κB p65 in the brain could regulate the expression of many genes, most of which encode immune proteins with pro-in ammatory activities [50].
Consistent with a previous study [51], we detected that cortical NF-κB p65 translocated from the cytoplasm to nuclei in the cortex in mice-treated with 60% HFD, indicating the activation of NF-κB p65 signaling pathway. While, the treatment of PUFAs rich HFD treatment inhibited the intracellular translocation of NF-κB p65, especially in n-3 PUFAs rich HFD-fed animals. We also found that APP/PS1 mice showed signi cantly up-regulated cortical TNF-α, COX2, IL-1β and IL-6 mRNA expressions, but lower iNOS mRNA expression than C57BL/6J mice, demonstrating an enhanced neuro-in ammatory response in these AD model mice. Anthony Pinçon et al. reported that administration of a high-cholesterol diet caused 38% increase of cortical TNF-α mRNA expression in APP/PS1 mice as compared with control mice without affecting IL-6 and IL-1β mRNA expressions [52]. Inconsistent with this report, we observed a signi cant decrease in cortical TNF-α, COX2 and IL-6 mRNA expression, but an increase in iNOS mRNA expression in HFD-fed APP/PS1 mice. The discrepancy between studies might attribute to the diversity of intervention diet or the study design. Moreover, we observed that the 60% and 45% HFDs did not affect cortical IL-1β mRNA expression. In contrast, n-3 PUFAs-rich 45% HFD signi cantly increased cortical IL-1β mRNA level, and a decrease expression was observed in mice treated with n-6 PUFAs-rich 45% HFD.
Although it is commonly recognized that n-6 PUFA has pro-in ammatory functions as prostaglandin and leukotriene precursors, n-6 PUFA can also be converted into anti-in ammatory mediators or modulate the gene expression of the pro-in ammatory cytokine to exert anti-in ammatory effects. Alashmali et al. reported that adequate dietary n-6 PUFA intake restored hippocampal in ammation cytokine gene expression caused by intra-cerebroventricular lipopolysaccharide (LPS) injection [53]. We, therefore, speculate that the AD-like pathology in APP/PS1 mice might have caused hyperactivated neuroin ammation in the brain, which could be antagonized by the treatment with HFD. The inconsistent expressional pattern of pro-in ammatory cytokines in the cortexes of HFD-fed mice further indicated the potentially varied regulatory effect(s) of dietary fatty acids on the cerebral in ammatory response. Thus, further studies are needed to elucidate the relationship between HFD (especially with different PUFAs ratios) and neuroin ammation in modulating AD pathology process.

Limitations
A few limitations remain in this study. First, due to the young age of the mice and the short duration of the intervention, no signi cant differences were found in learning and memory abilities. Therefore, a much longer period of dietary intervention was required of to explore the persistent effects of dietary PUFAs on these behavioral indices. Second, we only used male APP/PS1 mice for experiment, while recent studies have shown that APP/PS1 transgenic mice showed a gender difference in behavior and AD-like pathology in brain [54]. Thus, it is essential to further investigate the sex-based effects of dietary PUFAs on APP/PS1 mice. Also, due to the limited brain species, we fall to detect the level of proin ammatory cytokine in the brain, thus, we could not provide a more accurate cerebral neuroin ammatory pro le in the experimental animals.

Conclusion
In conclusion, HFD induced insulin resistance phenotype in peripheral tissues and increased circulating pro-in ammatory cytokine levels in APP/PS1 mice. Dietary n-6/n-3 PUFA ratio plays a critical role in modifying AD pathological process, insulin signaling pathway, and neuroin ammation in the cortex in response to high-fat diet treatment. In preventing AD, optimal dietary n-6/n-3 PUFAs ratio should be taken into consideration in PUFA-forti ed diet interventional strategy. Extensive studies are essentially needed to further uncover the relevant underlying molecular mechanisms and bridge the knowledge gap associated with the potential effects of dietary PUFAs on insulin resistance and cortical degeneration in AD. Availability of data and materials

Abbreviations
The datasets used and analyzed during the current study are available from thecorresponding author on reasonable request.
interpretation ofthe data; preparation, review, or approval of the manuscript; and decision tosubmit the manuscript for publication.