Necrostatin-1 Mitigates Cognitive Dysfunction in Prediabetic Rats With No Alteration in Insulin Sensitivity

Obesity is defined as an excess of adipose tissue accumulation, and it affects ∼39% of the world’s population (1). Obesity with impaired glucose tolerance and insulin sensitivity is known as a prediabetes condition. This condition increases the risk of several chronic diseases as well as neurodegeneration (2). Clinical data shows that prediabetes has a negative correlation with cognitive flexibility, fluid intelligence, working memory, visuospatial ability, and verbal memory (3). Our previous studies also reported cognitive impairment in rats with high-fat diet (HFD)-induced obesity and insulin resistance or rats with HFD-induced prediabetes (4,5). The mechanisms responsible for the cognitive dysfunction associated with the prediabetes condition have been shown to be involved with brain insulin receptor dysfunction (6), synaptic dysplasticity (6,7), brain mitochondrial dysfunction (6,7), and apoptotic cell death (4,5).

Among types of cell death, necroptosis has been described as programmed necrosis (8,9). It shares similar mechanisms and morphology with both apoptosis and necrosis (9). Necroptosis is confirmed by the activation of receptor-interacting protein (RIP) 1 and 3, cellular swelling, and cellular rounding (10). In 2005, Degterev et al. (9) identified a specific and potent small molecule that inhibits necroptosis, known as necrostatin-1 (nec-1). The efficacy of nec-1 was demonstrated on the brain in rodents with transient focal cerebral ischemia, showing that nec-1 effectively reduced brain infarct size compared with a vehicle group (9). In addition, nec-1 exerted beneficial effects on the brain in the case of several brain pathologies such as subarachnoid hemorrhage (11,12) and ischemic stroke (13,14). Nec-1 also reduced amyloid β (Aβ) aggregation in mice with Alzheimer disease (15,16) and improved cognitive function in mice with d-galactose–induced aging (17). However, the effects of nec-1 on brain pathology and cognition in prediabetic rats have not been investigated. In addition, metformin, an antidiabetes drug, showed beneficial effects on the recovery of brain function via improved brain mitochondrial function and attenuated brain oxidative stress, leading to improved cognition in prediabetic rats (7,18). Metformin is the drug of choice for patients with type 2 diabetes (19). Thus, in many experimental studies, metformin has been used as a standard treatment for comparison with other novel drugs or natural extracts in prediabetic and diabetic animal models (20–22). Previous studies demonstrated the beneficial effects of metformin in the brain including an improved cognitive function in diabetes and prediabetes conditions (18,20). Mechanistically, metformin at 300 mg/kg could reduce brain oxidative stress in both cortex and hippocampus (23,24). In addition, metformin could improve brain mitochondrial function as indicated by reducing brain mitochondrial oxidative stress, membrane potential changes, and swelling in prediabetic rats, leading to improved cognitive function (18). Therefore, metformin was selected as a standard treatment in the current study, and we aimed to compare the efficacy of nec-1 with that of metformin regarding metabolic parameters, cognitive function, and brain pathologies in prediabetic rats.

In this study, we hypothesized that nec-1 and metformin equally improve cognitive function in prediabetic rats through reduction of synaptic dysplasticity, microglial hyperactivity, brain mitochondrial dysfunction, brain inflammation, Alzheimer-related protein, and brain cell death.

All experimental protocols were approved by the Institutional Ethics Committee for Animal Research, Faculty of Medicine, Chiang Mai University (permit no. 28/2018). Thirty-two male rats were obtained from Nomura Siam Company (Bangkok, Thailand). Rats were randomly assigned into two dietary groups: a normal diet (ND, n = 8) and HFD (n = 24). ND-fed rats were given a standard chow diet containing 19.77% of energy from (E) from fat, and HFD-fed rats were fed a customized diet containing 59.28% of energy from fat (4,6,18). All rats were given their assigned diet throughout the experiment. Our previous studies showed that HFD-fed rats exhibited prediabetes after 12 weeks of HFD feeding (6,18); thus, all pharmacological interventions were commenced at week 13. At week 13, HFD-fed rats were divided into three subgroups: 1) vehicle (normal saline solution, with subcutaneous injection [s.c.]), 2) nec-1 (1.65 mg/kg/day s.c.), and 3) metformin (a positive control [300 mg/kg/day by oral gavage feeding]). Rats were given their assigned treatments for 8 weeks. The dosage of both nec-1 (1.65 mg/kg/day) and metformin (300 mg/kg/day) was selected based on results from previous reports (18,25).

After 8 weeks of treatment, cognitive function tests were performed. Blood was collected to determine metabolic parameters. An oral glucose tolerance test (OGTT) was used to determine peripheral insulin sensitivity. Finally, all rats were killed by decapitation. Brain tissue was used to determine hippocampal synaptic plasticity, brain insulin receptor function, dendritic spine density, microglial morphology, and brain mitochondrial function; also, protein expression analysis was carried out for Alzheimer protein, necroptosis, and apoptosis. The scheme of the experimental protocol is shown in Fig. 1A.

In this study, we used eight animals/group. The hippocampus from each group (n = 4/group) was used for extracellular recording. Other four hippocampi from each group were used to determine dendritic spine density (n = 4/group) and microglial morphology (n = 4/group) and for Western blot analysis. The cortex from each group (n = 8/group) was used for brain mitochondrial function, brain oxidative stress, and Western blot analysis in cortex. Fresh tissues were used for electrophysiological study and brain mitochondrial function study. For the electrophysiological study, the brain was immersed in ice-cold artificial cerebrospinal fluid (aCSF), containing NaCl 85 mmol/L, KCl 2.5 mmol/L, MgSO₄ 4 mmol/L, CaCl₂ 0.5 mmol/L, NaH₂PO₄ 1.25 mmol/L, NaHCO₃ 25 mmol/L, glucose 25 mmol/L, sucrose 75 mmol/L, kynurenic acid 2 mmol/L, and ascorbate 0.5 mmol/L, saturated with 95% O₂ and 5% CO₂ (pH 7.4). This solution helped to enhance neuronal survival during the slicing procedure. The hippocampal slices were cut using a vibratome (Vibratome Company, St. Louis, MO). Following a 30-min postslice incubation in high-sucrose aCSF, the slices were transferred to a standard aCSF solution, containing NaCl 119 mmol/L, KCl 2.5 mmol/L, CaCl₂ 2.5 mmol/L, MgSO₄ 1.3 mmol/L, NaH₂PO₄ 1 mmol/L, NaHCO₃ 26 mmol/L, and glucose 10 mmol/L, and saturated with 95% O₂/5% CO₂ (pH 7.4) for an additional 30 min at room temperature (22–24°C) (6) (n = 4 animals/group and 1–2 brain slices/each animal).

For brain mitochondrial function study, after decapitation, the cortex was placed and homogenized in 10 mL cold mannitol-glucose-ethylenen glycol tetraacetic acid (MSE) solution containing mannitol 225 mmol/L, sucrose 75 mmol/L, EGTA 1 mmol/L, HEPES 5 mmol/L, and BSA 1 mg/mL, pH 7.4 (18).

For Western blot analysis in the cortex, the cortex was rapidly frozen in liquid nitrogen and kept in −85°C until analysis. For Western blot analysis in the hippocampus tissue, a separate set of animals (n = 4) was used, hippocampus was dissected on ice, and the tissue was rapidly frozen in liquid nitrogen and kept in −85°C until analysis. For Golgi staining and microglial morphology, the brain was rapidly fixed in either the solution A + B of the FD Rapid GolgiStain Kit (FD NeuroTechnologies, Columbia, MD) to determine dendritic spine density or 10% formalin to determine microglial morphology (n = 4 animals/each group and 2–4 brain slices/each animal). The diagram of brain sections used in this study is shown in the Fig. 1B.

Fasting plasma glucose, total cholesterol, triglyceride (Erba Diagnostics, Mannheim, Germany), and HDL levels (BioVision, Inc.) were measured using commercial colorimetric kits. The Friedewald formula was used to analyze LDL levels. The degree of insulin resistance was calculated by HOMA of insulin resistance (18).

Plasma and brain malondialdehyde (MDA) levels were used to assess systemic and brain oxidative stress, respectively. MDA levels were detected by measuring the absorbance of thiobarbituric acid–reactive substances with a high-performance liquid chromatography method as previously described (18).

The novel object recognition (NOR) test is a hippocampal-independent learning and memory test. The NOR was conducted first at day 1 and day 2, and the rats were rested for 5 days (days 3–7). Then, the novel object location (NOL) test was performed at days 8–10. The habituation phase was performed on day 1. Rats freely explored the box with no objects for 10 min, and the locomotor activity was analyzed. The familiarization phase was assessed at day 2. Each rat was placed in a box with two similar objects for 10 min, and the percentage of exploration time was recorded in the next 5 min. A video camera mounted on the wall directly above the box was used to record the testing session for off-line analysis. The time of exploration was manually counted. Percentage index preference was calculated from the following formula: % index preference = [(time on object 2) ÷ (time on object 1 + time on object 2)] × 100 (26).

The NOL test is a hippocampal-dependent learning and memory test. For the habituation phase, each rat was placed into the circle box for 10 min. After 24 h of a habituation phase, a familiarization phase was performed. Each rat was placed again in the center of the apparatus and left to explore two similar objects for 10 min. On the following day, rats were placed in the same box, which contained two familiar objects; however, one object was changed to a new location. The test was performed for 10 min. During each phase, a video was recorded with a recording camera above the box for off-line analysis. The percentage exploration time and percentage preference index were calculated from the following formula: % index preference = [(time on object with new location) ÷ (time on object 1 + time on object with new location)] × 100 (26).

Brain slices were transferred to a submersion recording chamber and continuously perfused at 3–4 mL/min with standard aCSF warmed to 28–29°C. Field excitatory postsynaptic potentials (fEPSPs) were stimulated with a bipolar tungsten electrode at the Schaffer collateral-commissural pathway, while the fEPSP tracing was recorded at the stratum radiatum of the hippocampal CA1 region (6).

For high-frequency–induced long-term potentiation (LTP), a high-frequency stimulation (100 Hz, 0.5 s duration, 20 s interval) was delivered at 1.5 times the baseline stimulation intensity. The LTP tracing was recorded for at least 50 min after high-frequency stimulation. The amount of potentiation was calculated at 50 min after tetanus (7).

For insulin-induced long-term depression (LTD), the brain slices were stimulated using low frequency: 0.033 Hz. The brain slices were perfused with aCSF for 10 min as baseline and changed to aCSF plus 500 nmol/L insulin for an additional 10 min. The slices were perfused with aCSF again for a further 50 min, and results were recorded. Both LTP and LTD tracing were filtered at 3 kHz. They were digitized at 10 kHz and analyzed using pCLAMP 9.2 software (Axon Instruments). The initial slope of the fEPSPs was measured and plotted against time (6).

The brain was fixed with a commercial Golgi staining solution (FD NeuroTechnologies) in a dark room and stored in 4°C for 2 weeks, and it was then cut into 60-μm-thick slices using Cryostat (Lieca CM1950; Leica Biosystems Nussloch GmbH, Nussloch, Germany). The dendritic spines were visualized under a microscope (IX-81; Olympus, Tokyo, Japan), and the number of spines were counted from 20 nm of the apical end of the dendrite by using xcellence software (Olympus) (27).

After decapitation, the brain tissues were fixed with 4% paraformaldehyde for 24 h, cryoprotected in 30% sucrose in PBS at 4°C, and then frozen in isopentane and dry ice and stored at −80°C. The brain was cut into 20-μm sections by using cryosection (Leica CM1950; Leica Biosystems Nussloch GmbH). The sections were incubated with anti–Iba-1 (Abcam, Cambridge, U.K.) and costained with DAPI (Abcam) for 12 h. After being washed three times in TBS, sections were incubated with Alexa Fluor–conjugated secondary antibody for 1 h at 25°C and treated with copper sulfate in ammonium acetate buffer to quench endogenous autofluorescence of the brain tissue. Then, the series of z-stacks of microglia images was captured using fluorescent microscopy (Nikon, Tokyo, Japan) and microglial morphology parameters were measured by Imaris software 7.0 (Bitplane, Zurich, Switzerland) (27).

Brain mitochondria were obtained by a differential centrifugation technique as previously described (18). In brief, after the brain was removed, it was placed into 5 mL ice-cold MSE solution (225 mmol/L mannitol, 75 mmol/L sucrose, 1 mmol/L EGTA, 5 mmol/L HEPES, and 1 mg/mL BSA, pH 7.4) to rapidly wash out the blood. Then, the brain was transferred to 10 mL ice-cold MSE solution containing 0.05% proteinase, bacterial (Sigma-Aldrich), and homogenized by use of a homogenizer at 600 rpm. The homogenate was centrifuged at 2,000g for 4 min, and the supernatant was collected. After that, the supernatant was centrifuged at 12,000g for 9 min. The mitochondrial pellets were collected and resuspended in 4 mL ice-cold MSE-digitonin solution (0.02% digitonin in MSE solution). The solution was centrifuged at 12,000g for another 11 min (18). Finally, the mitochondrial pellet was collected and resuspended in a respiration buffer containing 150 mmol/L KCl, 5 mmol/L HEPES, 5 mmol/L K₂HPO₄0.3H₂O, 5 mmol/L l-glutamate, and 5 mmol/L pyruvate sodium salt. Mitochondrial protein was determined by the bicinchoninic acid assay. Isolated brain mitochondria at the concentration 0.4 mg/mL was used to determine mitochondrial function including brain mitochondrial reactive oxygen species (ROS) levels, brain mitochondrial membrane potential changes, and brain mitochondrial swelling. Brain mitochondrial ROS levels were determined using the fluorescence intensity of dichlorofluorescein (DCF). Brain mitochondrial membrane potential changes were determined using a red-to-green fluorescence intensity ratio of tetraethylbenzimidazolylcarbocyanine iodide (JC1). For both protocols for measurement of mitochondrial ROS and mitochondrial membrane potential changes, 2 mmol/L H₂O₂ was added to mimic a severe oxidative stress condition, and the percentage changes of fluorescence intensity between mitochondria with and without H₂O₂ was calculated. An increase in percentage change indicated brain mitochondrial dysfunction. Brain mitochondrial swelling was determined by the absorbance of mitochondrial protein in a respiration buffer. All measurements were made using a fluorescent/absorbance microplate reader (BioTek). Representative pictures of brain mitochondria were taken using a transmission electron microscope (Jeol, Tokyo, Japan).

The cortex and hippocampus were lysed with extraction buffer. The cortex lysate (1.8–2.7 mg/mL) or hippocampal lysate (0.9 mg/mL) was used to determine the protein expression by Western blot analysis. The tissue lysate was separated by gel electrophoresis on 10% polyacrylamide gel and then transferred onto nitrocellulose membranes. Immunoblots were blocked for 1 h with 5% nonfat dry milk or BSA in Tris-buffer saline (pH 7.4) containing 0.1% Tween 20. Then, the membranes were probed with anti-claudin 5 (Abcam), p-NFkB (Cell Signaling Technologies), NFkB (Cell Signaling Technologies), p-RIP1 (Cell Signaling Technologies), RIP1 (Cell Signaling Technologies), RIP3 (Cell Signaling Technologies), cleaved caspase 3 (Cell Signaling Technologies), caspase 3 (Cell Signaling Technologies), Bax (Cell Signaling Technologies), Bcl2 (Abcam), p-Tau (Cell Signaling Technologies), Tau (Cell Signaling Technologies), and actin (Santa Cruz Biotechnology). The assessment of protein density was carried out using a ChemiDoc touching system (Bio-Rad), and the densitometric analysis was performed by ImageJ (National Institutes of Health) (27).

Data are expressed as mean ± SEM, and data were processed using GraphPad Prism software (version 7; GraphPad Software, Inc.). A one-way ANOVA followed by a least significant differences post hoc test was used to test the differences between groups. P < 0.05 was considered to indicate statistical significance.

The data sets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request. No applicable resources were generated or analyzed during the current study.

HFD-fed rats treated with vehicle exhibited obesity and insulin resistance, as indicated by body weight gain (P = 0.003), increased visceral fat weight (P < 0.0001), increased fasting plasma insulin levels (P = 0.0251), and HOMA index (P = 0.0201) (Table 1). The area under the curve (AUC) of plasma glucose levels during OGTT was increased in HFD-fed rats treated with vehicle compared with that of ND-fed rats with vehicle (P = 0.0185) (Table 1). Plasma total cholesterol (P = 0.0159) and LDL levels (P = 0.0014) were significantly increased, while plasma HDL level was not changed in HFD-fed rats treated with vehicle compared with ND-fed rats treated with vehicle (P = 0.0932) (Table 1). An increase in plasma MDA levels was observed in HFD-fed rats treated with vehicle (P = 0.0150) (Table 1). These data suggest that HFD-fed rats developed obesity, peripheral insulin resistance, dyslipidemia, and increased systemic oxidative stress.

Treatment with metformin effectively reduced body weight (P = 0.003), visceral fat weight (P = 0.0007), plasma insulin (P = 0.0033), HOMA index (P = 0.0021), AUC of plasma glucose during OGTT (P = 0.0449), fasting total cholesterol (P = 0.0069), fasting LDL (P = 0.0096), and MDA levels (P = 0.0003) in HFD-fed rats (Table 1). Treatment with nec-1 significantly reduced visceral fat weight (P = 0.0014) and serum MDA levels (P = 0.0008) compared with HFD-fed rats treated with vehicle (Table 1). However, nec-1 did not affect body weight, fasting plasma insulin, HOMA index, AUC of plasma glucose during OGTT, plasma cholesterol, or plasma LDL levels compared with HFD-fed rats treated with vehicle (Table 1). These data indicate that nec-1 reduced visceral fat deposition and peripheral oxidative stress, but it did not result in any improvement in insulin sensitivity in prediabetic rats.

Hippocampal-dependent learning and memory behavior was tested using the NOL task (Fig. 2A) and showed that % preference of two objects during the familiarization phase was not different between groups (Fig. 2B). However, the % preference of two objects during the testing phase was decreased in HFD-fed rats treated with vehicle (P < 0.0001), and it was restored back to the normal levels in HFD-fed rats treated with both metformin and nec-1, compared with that of ND-fed rats (P = 0.6760, P = 0.2159, respectively) (Fig. 2C).

Hippocampal-independent learning and memory behavior was tested using the NOR task (Fig. 2D). Our results showed that % preference of two objects was not different between groups during the familiarization phase (Fig. 2E); however, the % preference of two objects was decreased in HFD-fed rats treated with vehicle compared with ND-fed rats treated with vehicle (P = 0.0007) (Fig. 2F). Treatment with both metformin and nec-1 equally restored % preference of two objects during the testing phase to the same level as in ND-fed rats (P = 0.6183, P = 0.6227, respectively) (Fig. 2F). These data suggested that HFD-fed rats developed impairment of both hippocampal-dependent and -independent learning and memory. Data suggested that treatment with metformin and treatment with nec-1 shared similar efficacy in improving hippocampal-dependent and -independent learning and memory in prediabetic rats.

Hippocampal synaptic plasticity plays a vital role in regulating learning and memory. Data from a high-frequency–stimulated LTP demonstrated that both the % normalized fEPSP slope and % increment of the fEPSP slope were decreased in HFD-fed rats treated with vehicle (P < 0.0001), and it was restored back to the normal level in HFD-fed rats treated with metformin and nec-1, when compared with that of ND-fed rats (P = 0.6997, P = 0.8789, respectively) (Fig. 3A and B). In addition, brain insulin receptor function was determined using an insulin-induced LTD protocol. The results showed that both the % normalized fEPSP slope and % decrement of the fEPSP slope were decreased in HFD-fed rats treated with vehicle compared with those of ND-fed rats (P < 0.0001). Treatment with metformin alone restored the % normalized fEPSP slope and % decrement of fEPSP slope of LTD in HFD-fed rats, when compared with those of ND-fed rats (P = 0.6997) (Fig. 3C and D). Taken together, our data suggested that HFD consumption led to impairment of brain insulin receptors and hippocampal synaptic plasticity. Both metformin and nec-1 improved hippocampal synaptic plasticity to the same extent; however, only metformin improved brain insulin receptor function in prediabetic rats. These findings suggested that the beneficial effect of nec-1 on hippocampal synaptic plasticity was not associated with brain insulin receptor function.

Therefore, we further investigated the underlying mechanism responsible for the beneficial effects of nec-1 on learning and memory. We found that the dendritic spine density was significantly decreased in HFD-rats treated with vehicle (P < 0.0001), and treatment with both metformin and nec-1 significantly alleviated the reduction of dendritic spine density in the CA1 region of the hippocampus (P < 0.0001, P = 0.0002, respectively) (Fig. 4A). In addition, claudin 5 protein levels were decreased in the hippocampus of HFD-fed rats treated with vehicle (P = 0.0045). Treatment with both metformin (P = 0.0112) and nec-1 (P = 0.0081) similarly increased claudin 5 protein levels in HFD-fed rats (Fig. 4B). These data suggested that HFD consumption reduced hippocampal dendritic spine density and blood-brain barrier proteins. Metformin and nec-1 potentially alleviated the reduction of hippocampal dendritic spine density and blood-brain barrier disruption in prediabetic rats.

An alteration in microglial morphology was determined in the hippocampal region of rats. Iba-1 was used as a marker of microglia. The representative images of microglia are shown in Fig. 5A. The results showed that number and volume of Iba-1–positive cells did not differ between groups (Fig. 5B and C). We found that Iba-1 process length (P < 0.0001) and AUC of Iba-1 intersections (P < 0.0001) were significantly lower in HFD-fed rats treated with vehicle than those in ND-fed rats treated with vehicle (Fig. 5D–F). Treatment with both metformin and nec-1 significantly improved Iba-1 process length (P < 0.0001 and P < 0.0001, respectively) and AUC of Iba-1 intersections (P < 0.0001 and P < 0.0001) in HFD-fed rats. These data suggested that HFD consumption increased microglial hyperactivity in the hippocampus, and treatment with metformin and nec-1 reduced microglial hyperactivity to an equal degree in the hippocampus of prediabetic rats.

Our data showed that brain MDA levels, representing brain oxidative stress, were increased in HFD-fed rats treated with vehicle compared with those of ND-fed rats (P = 0.0284) (Fig. 6A). Treatment with metformin and nec-1 equally reduced brain MDA levels in HFD-fed rats compared with HFD-fed rats treated with vehicle (P = 0.0157, P = 0.0082, respectively) (Fig. 6A). Mitochondria are known as being the major source of oxidative stress production and energy production sites in the brain. Data from brain mitochondrial function showed that HFD-fed rats treated with vehicle had a higher degree of mitochondrial ROS (P < 0.0001), mitochondrial membrane depolarization (P = 0.0025), and mitochondrial swelling (P < 0.0001) (Fig. 6B–D). Treatment with both metformin and nec-1 effectively restored brain mitochondrial function back to normal levels via reducing mitochondrial ROS levels (P < 0.0001 and P < 0.0001), mitochondrial membrane depolarization (P = 0.0141 and P = 0.0156), and mitochondrial swelling (P < 0.0001 and P < 0.0001) (Fig. 6B–D). The representative images of brain mitochondria from the transmission electron microscope are shown in Fig. 6E. These data indicated that HFD consumption increased brain oxidative stress and mitochondrial dysfunction, and treatment with metformin and nec-1 mitigated brain oxidative stress and mitochondrial dysfunction in prediabetic rats.

The expression of p-RIP1 protein was determined in both hippocampus and cortex. Our results showed that the HFD-fed rats had higher p-RIP1 levels in both the hippocampus (P < 0.0001) and the cortex (P < 0.0001), when compared with ND-fed rats treated with vehicle, and the level of necroptosis was not different between these two regions in HFD-fed rats. Treatment with metformin and nec-1 equally reduced p-RIP1 level in both hippocampus (P = 0.0012 and P = 0.0004, respectively) and cortex (P < 0.0001 and P < 0.0001) compared with levels in HFD-fed rats treated with vehicle (Supplementary Fig. 1). Moreover, the levels of p-RIP1 were not different between these two regions (Supplementary Fig. 1). Therefore, these data suggested that metformin and nec-1 exerted similar effects on reducing necroptosis in the hippocampus and the cortex of prediabetic rats (Supplementary Fig. 1).

Cortex lysates were used to determine the expression of several proteins, such as inflammatory proteins (p-NFkB/NFkB), necroptotic proteins (p-RIP1/RIP1 and RIP3), apoptotic proteins (cleaved caspase 3/caspase 3, Bax/Bcl2), and Alzheimer-related protein (p-Tau/Tau). The results showed that the expressions of proteins p-NFkB/NFkB (P = 0.0005), p-RIP1/RIP1 (P < 0.0001), RIP3 (P = 0.0213), and p-Tau/Tau (P = 0.0176) were significantly increased in HFD-fed rats treated with vehicle compared with those in ND-fed rats (Fig. 7A–C and F). However, the expressions of cleaved caspase 3/caspase 3 and Bax/Bcl2 were not altered in HFD-fed rats treated with vehicle compared with ND-fed rats (Fig. 7D–E). These findings suggested that HFD consumption led to brain inflammation, necroptosis, and Tau hyperphosphorylation. However, brain apoptosis was not observed in prediabetic rats. Both treatment with metformin and treatment with nec-1 effectively reduced p-NFkB/NFkB (P = 0.0406 and P = 0.0011, respectively), p-RIP1/RIP1 (P < 0.0001 and P = 0.0001), p-RIP3 (P = 0.0134 and P = 0.0029), and p-Tau/Tau (P = 0.0244 and P = 0.0407) levels in HFD-fed rats compared with levels in HFD-fed rats treated with vehicle (Fig. 7A–C and F). These data suggested that both metformin and nec-1 equally reduced brain inflammation, necroptosis, and hyperphosphorylated Tau in prediabetic rats.

The protein expressions of p-IRS1^Ser307, IRS1, p-Akt^Ser473, and Akt were decreased in HFD-fed rats treated with vehicle compared with those of ND-fed rats (Supplementary Fig. 2). Metformin treatment increased the protein expressions of p-IRS1^Ser307, IRS1, p-Akt^Ser473, and Akt in HFD-fed rats compared with those in HFD-fed rats treated with vehicle (Supplementary Fig. 2). However, nec-1 did not affect these protein expressions in liver. These findings confirmed that nec-1 did not improve peripheral insulin sensitivity, whereas metformin did.

The major findings of this study are that 1) long-term HFD consumption leads to a prediabetes condition and cognitive decline with several brain dysfunctions, including brain synaptic dysplasticity, inflammation, microglial hyperactivity, mitochondrial dysfunction, necroptosis, and increased levels of Alzheimer-related protein; 2) metformin effectively reduced obesity, dyslipidemia, and peripheral and brain insulin resistance in prediabetic rats; 3) nec-1 reduced visceral fat to an extent similar to that seen with metformin, but it did not reduce peripheral or brain insulin resistance or dyslipidemia in prediabetic rats; 4) both nec-1 and metformin showed similar efficacy in improving cognitive function in both hippocampal-dependent and -independent manners through reducing brain inflammation, microglial hyperactivity, brain mitochondrial dysfunction, necroptosis, and Alzheimer-related protein levels.

In this study, we show that nec-1 improved cognitive function in male prediabetic rats. Although there is evidence that diabetes may confer a greater risk of developing dementia for women than for men (28) and female rats have been shown to not inherently contain more variability than male rats (29,30), the meta-analyses in the general population have demonstrated that there are no differences in terms of behavior, electrophysiology, histology, neurochemistry, and nonbrain measures between male and female rats (29,30) However, the characteristics of male prediabetic rats were different from those of female prediabetic rats. For example, 1) body weight: male prediabetic rats were 65% heavier than that female prediabetic rats; 2) plasma insulin levels of male prediabetic rats were 121% greater than those of female rats; and 3) HOMA index of male prediabetic rats was 93% greater than that of female prediabetic rats (31). Because of many differences in the metabolic parameters between male and female prediabetic rats in which the systemic effects were more severe in male than in female rats, male prediabetic rats were chosen as a study model in this study. Future study is needed to investigate the roles of necroptosis as well as the effects of nec-1 on brain pathology in female prediabetic rats.

Several previous studies have shown that long-term HFD consumption induces a prediabetes condition with brain insulin resistance and cognitive decline (18,32,33). The potential mechanisms associated with cognitive decline in prediabetic rats are brain oxidative stress, mitochondrial dysfunction, inflammation, insulin resistance, microglial hyperactivity, synaptic dysplasticity, and cell death. A previous study also showed that long-term HFD consumption also increased apoptotic markers in hippocampus of female rats, as indicated by increasing levels of cleaved caspase 3 and Bax (34). However, our data showed that cleaved caspase 3/caspase 3 and Bax/Bcl-2 in the cortex were not different between the ND- and HFD-fed male rats. Interestingly, the necroptotic markers including p-RIP1/RIP1 and p-RIP3 were upregulated in the brain of the prediabetic rats in this study. These data suggest that the prediabetes condition leads to cognitive decline via several brain dysfunctions as well as brain cell death through necroptosis but not apoptosis.

Microglia are resident immune cells in the brain, and they are responsible for brain inflammation. Microglia can be divided into two forms: amoeboid shape and ramified shape. Normally, the ramified shape is represented as being the steady state of microglia and has multiple branches and long process length. When inflammation increases, microglia are triggered to change from the ramified shape to an amoeboid shape by enlargement of the cell body and shortening of process length. The amoeboid morphology represents an activated state of microglia, which is associated with a proinflammatory function (35). Microglial hyperactivity and an increase in NFkB levels were observed in brain of prediabetic rats, suggesting that brain inflammation occurred. Several studies demonstrated that inflammation and oxidative stress in the brain suppressed dendritic spine formation (36) and increased Tau hyperphosphorylation (37,38). Those findings suggest that the reduction of dendritic spine density synaptic dysplasticity, and increased Tau hyperphosphorylation in prediabetic rats, could be due to not only microglial hyperactivity but also an increase in brain oxidative stress and inflammation.

Treatment with nec-1 effectively reduced brain inflammation to an extent similar to that seen with metformin treatment as indicated by a reduction in NFkB activation. A previous study has shown a relationship between NFkB and microglial activation (39). Pathological stress can activate NFkB activation, which further transforms resident microglia to its amoeboid shape (39). Therefore, a reduction of NFkB function by nec-1 and metformin leads to a reduction in microglial hyperactivity and a reduction in Tau hyperphosphorylation in prediabetic rats.

In addition, nec-1 attenuated not only microglial dysmorphology but also brain mitochondrial dysfunction and oxidative stress. A previous study reported that brain mitochondrial dysfunction was observed in the obese condition and resulted in cognitive decline (18). Mitochondria play an important role in regulating energy balance and cellular function. In the central nervous system, several processes of the brain are regulated by mitochondria. These include synaptic transmission and neurogenesis, which are key mechanisms for learning and memory (40). One study has reported that mitochondrial dysfunction has been associated with neurological disorders and cognitive impairment (41). The beneficial effects of nec-1 on microglia and mitochondria, as well as the reduction in brain inflammation and Tau phosphorylation, could lead to a reduction in dendritic spine loss and improved synaptic plasticity. Those effects of nec-1 led to alleviated cognitive decline in prediabetic rats.

Our data demonstrated that nec-1 did not affect peripheral and brain insulin sensitivity in prediabetic rats, but it partially reduced total cholesterol and LDL levels. This finding was inconsistent with findings in a previous report, in which Xu et al. (42) reported that nec-1 (2 mg/kg/day for 4 weeks) reduced insulin resistance in ob/ob mice via improvement of insulin sensitivity and decrease in hepatic triglyceride levels. The controversy about these data could be explained by the differences in species, dose of nec-1, and duration of treatment. Therefore, further study should investigate the impact of a higher dose of nec-1 in the brain in a prediabetic condition.

Our data demonstrated that treatment with nec-1 resulted in the inhibition of RIPK1 activity, leading to reduced necroptosis and improved cognitive function in prediabetic rats. Also, our results showed that nec-1 did not alter brain insulin resistance in these prediabetic rats, when compared with prediabetic rats treated with vehicle, as indicated by the absence of an insulin-induced LTD phenomenon in prediabetic rats treated with nec-1. Moreover, nec-1 did not increase insulin signaling function including p-IRS1 and p-Akt in liver of prediabetic rats. All of these findings indicated that the beneficial effect of nec-1 on the brain and cognition was independent of the improvement of brain insulin sensitivity.

The current study is the first study to show that the inhibition of necroptosis, by nec-1, in prediabetic rats directly reduces brain inflammation, microglial hyperactivity, brain mitochondrial dysfunction, and synaptic dysplasticity, leading to improved hippocampal-dependent and -independent learning and memory without alleviating the obese–insulin resistant condition. In comparison with metformin, which is a standard treatment in diabetes, nec-1 showed similar efficacy in reducing cognitive dysfunction, synaptic dysplasticity, oxidative stress, brain inflammation, mitochondrial dysfunction, and Alzheimer protein levels. However, nec-1 could not reduce peripheral or brain insulin resistance. In clinical studies, it has been shown that metformin, as a monotherapy, may not be adequate for patients with advanced stages of insulin resistance (the failure rate was 12–15%) (43,44). Therefore, our findings from this basic study suggested that nec-1 could be a novel therapeutic of choice or it could be used as an add-on treatment to metformin to attenuate prediabetes-induced cognitive decline.

Metformin therapy has showed several favorable outcomes regarding cognitive function in a prediabetes condition (33,34). In comparison with the impact of nec-1, our findings suggested that metformin and nec-1 equally improved cognitive function in both hippocampal-dependent and -independent learning and memory of prediabetic rats. Interestingly, we found that metformin can inhibit RIP1 and RIP3 phosphorylation in the brain of prediabetic rats. Adding weight to the findings of the current study, a previous study reported that metformin could inhibit necroptosis through the AMPK-Parkin axis by promoting polyubiquitination of RIP3, leading to the prevention of necrosome formation, which contributed to necroptosis (45). All of our findings suggest that metformin may have more extensive beneficial effects on the brain of prediabetic rats than nec-1 because metformin can alleviate brain pathologies and further increase peripheral and brain insulin sensitivity in this animal model. Regarding the mechanistic effects of metformin and nec-1 on cognitive function, the benefits of both drugs could be mainly due to the reduction of necroptosis regardless of insulin sensitivity effects. However, it is also possible that metformin could initially improve peripheral insulin sensitivity, after which the reduction of necroptosis occurred, leading to reduced cognitive decline in prediabetic rats. Future study is needed to investigate this proposed mechanism.

In summary, chronic HFD consumption led to a prediabetic condition and cognitive decline via causing several brain pathologies including insulin resistance, inflammation, microglial hyperactivity, mitochondrial dysfunction, oxidative stress, and necroptosis. Treatment with nec-1 directly improved cognition in the prediabetic condition through inhibiting necroptosis and reducing brain inflammation and mitochondrial dysfunction without an improvement in insulin sensitivity. However, metformin therapy in the prediabetic condition alleviated cognitive decline, possibly via both improving insulin sensitivity and reducing brain pathologies. Therefore, these findings suggest that metformin could have greater efficacy on cognition and brain functions than nec-1 in the prediabetic condition.

Funding. This work was supported by a Senior Research Scholar grant from the National Research Council of Thailand (to S.C.C.), the National Science and Technology Development Agency Thailand (NSTDA Research Chair grant to N.C.), the Chiang Mai University Center of Excellence Award (N.C.), and Thailand Research Fund grants TRG6280005 (to N.A.) and RTA 6080003 (to S.C.C.).

Duality of Interest. No potential conflicts of interest relevant to this article were reported.

Author Contributions. S.C.C. and N.C. designed the experiments. K.J., N.A., and S.W. conducted the experiments. K.J., N.A., W.P., N.C., and S.C.C. analyzed the data. K.J., N.A., N.C., and S.C.C. wrote the manuscript and finalized the manuscript. S.C.C. 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.