Amelioration of Hippocampal Insulin Resistance Reduces Tau Hyperphosphorylation and Cognitive Decline Induced by Isoflurane in Mice

General anesthetics can induce cognitive impairments and increase the risk of Alzheimer’s disease (AD). However, the underlying mechanisms are still unknown. Our previous studies shown that long-term isoflurane exposure induced peripheral and central insulin resistance (IR) in adult mice and aggravated IR in type 2 diabetes mellitus (T2DM) mice. Clinical and preclinical studies revealed an association between impaired insulin signaling and tau pathology in AD and other tauopathies. We investigated if alleviation of hippocampal IR by the antidiabetic agent metformin could reduce tau hyperphosphorylation and cognitive decline induced by isoflurane in mice. The effects of prolonged (6 h) isoflurane anesthesia on hippocampal IR, hippocampal tau hyperphosphorylation, and hippocampus-dependent cognitive function were evaluated in wild type (WT) adult mice and the high-fat diet plus streptozotocin (HFD/STZ) mouse model of T2DM. Here we shown that isoflurane and HFD/STZ dramatically and synergistically induced hippocampal IR and fear memory impairment. Metformin pretreatment strongly ameliorated hippocampal IR and cognitive dysfunction caused by isoflurane in WT mice, but was less effective in T2DM mice. Isoflurane also induced hippocampal tau hyperphosphorylation and metformin reversed this effect. In addition, isoflurane significantly increased blood glucose levels in both adult and T2DM mice, and metformin reversed this effect as well. Administration of 25% glucose to metformin-pretreated mice induced hyperglycemia, but surprisingly did not reverse the benefits of metformin on hippocampal insulin signaling and fear memory following isoflurane anesthesia. Our findings show hippocampal IR and tau hyperphosphorylation contribute to acute isoflurane-induced cognitive dysfunction. Brief metformin treatment can mitigate these effects through a mechanism independent of glycemic control. Future studies are needed to investigate whether long-term metformin treatment can also prevent T2DM-induced hippocampal IR and cognitive decline.


INTRODUCTION
About 10-50% of patients receiving surgery under anesthesia demonstrate cognitive impairments occurring between 1 month and 1 year after surgery following treatment, termed postoperative cognitive dysfunction (POCD) (James, 2020). POCD is not a transient phenomenon but associated with prolonged hospitalization, functional decline, less likely to live independently, and increased mortality (Moller et al., 1998;Steinmetz et al., 2009). Intraoperative hypoxia, inflammation, and the pharmacological effects of anesthetics are implicated in POCD pathogenesis, and older age is the strongest risk factor (Moller et al., 1998). Further, accumulating clinical and experimental evidence suggest that surgery under general anesthesia increases the risk of developing Alzheimer's disease (AD) and other forms of dementia (James, 2020). Our previous studies confirmed that long-term isoflurane exposure can induce cognitive impairment in adult mice, providing an experimental model to investigate the underlying pathogenesis (Xia et al., 2016;Song et al., 2019).
In addition to increasing POCD risk, aging increases susceptibility to metabolic syndrome (MetS) and type 2 diabetes mellitus (T2DM) (Ford et al., 2002;Menke et al., 2015). It is also well accepted that MetS can induce cognitive impairments, including of memory and executive function Ng et al., 2016). Moreover, MetS increases the risk of POCD (Hudetz et al., 2011). A decline in brain tissue volume can be detected even in early onset MetS patients (Sala et al., 2014). Similarly, T2DM disrupts brain structure and function, especially in cognition-related regions, and increases the risks of both POCD and age-related cognitive decline compared to age-matched non-diabetics (Feinkohl et al., 2017;Sanjari Moghaddam et al., 2019). Our pilot study also found that T2DM predicts POCD at 1 week after orthopedic surgery in elderly patients (over 60 years of age) and patients with T2DM had a 2.6-fold higher risk of POCD compared with patients without diabetes at 1 week after surgery (Supplementary Table 1). Collectively, these observations suggest that the molecular signaling mechanisms disrupted in T2DM may also contribute to POCD.
A decrease in insulin-induced molecular signaling, termed insulin resistance (IR), is a major pathogenic mechanism in MetS and T2DM, and may contribute to the associated cognitive impairments (Biessels and Reagan, 2015;. Experimental models of diabetes not only show remarkable peripheral and central nervous system IR, but also deficits in the neuroplastic processes linked to cognition (Biessels and Reagan, 2015). Moreover, brain and peripheral IR increase with advanced age (Petersen et al., 2015;Martín-Segura et al., 2019). Insulin resistance is also a common complication after Abbreviations: AD, Alzheimer's disease; Anes Isoflurane anesthesia; DM T2DM model mice; Glu 25% glucose solution; HFD, High-fat diet; IR, Insulin resistance; IRS1, Insulin receptor substrates 1; IRS2, Insulin receptor substrates 2; MCI, Mild cognitive impairment; Met Metformin injection; MetS, Metabolic syndrome; POCD, Postoperative cognitive dysfunction; STZ, Streptozotocin; T2DM, Type 2 diabetes mellitus; WT, Wild type. anesthesia and surgery, and the degree of postoperative IR is associated with surgical magnitude (Thorell et al., 1993). In aged rats as well, surgical procedures impair central insulin signaling and increase susceptibility to hippocampus-related cognitive impairments (Kawano et al., 2016). Isoflurane and sevoflurane reduced peripheral insulin sensitivity by about 50% and almost completely suppressed liver insulin responses in a canine model (Kim et al., 2016). These anesthetics also significantly reduced mean cerebral glucose utilization during anesthesia in rats (Lenz et al., 1998). Recently, we found that long-term isoflurane exposure induced peripheral and central IR in adult mice and aggravated IR in T2DM mice (Fang et al., 2020). Clinical studies show that preoperative IR predicts POCD in elderly gastrointestinal patients and that the IR index is significantly higher in patients with POCD than patients without POCD (Tang et al., 2017;He et al., 2019). Collectively, these results provide compelling evidence that IR contributes to POCD development.
Hippocampal accumulation of hyperphosphorylated tau protein may link MetS/T2DM, IR, POCD, and AD risk. Hyperphosphorylated tau proteins are a core component of neurofibrillary tangles, a pathological hallmark of AD (Ittner et al., 2010), and IR can prevent hippocampal tau dephosphorylation in rats . Both clinical and preclinical studies have found an association between impaired insulin signaling and tau pathology in AD and other tauopathies (Yarchoan et al., 2014). Further, inhalational anesthetic exposure, especially prolonged or repeated exposure, induced a long-lasting increase in tau phosphorylation associated with cognitive impairment (Dong et al., 2012;Le Freche et al., 2012). Conversely, insulin-sensitizing strategies have been shown to improve memory performance, reduce cerebrospinal fluid biomarkers of disease, and enhance cerebral glucose utilization in mild cognitive impairment (MCI) and AD patients (Hölscher, 2014).
Based on these findings, we speculated that isofluraneinduced hippocampal IR and ensuing tau hyperphosphorylation contribute to POCD and progression of AD pathology, while amelioration of hippocampal IR may mitigate tau hyperphosphorylation and POCD caused by long-term isoflurane anesthesia. Metformin, a first-line antidiabetic drug, exerts its therapeutic glucose-lowering effects by suppressing hepatic gluconeogenesis, decreasing insulin resistance, and increasing insulin sensitivity (Giannarelli et al., 2003). Here we examine whether pretreatment with metformin can alleviate hippocampal IR and tau hyperphosphorylation as well as cognitive impairment caused by long-term isoflurane anesthesia in adult and T2DM mice.

Animals and Treatment
Animal care and study protocols were approved by the Laboratory Animal Ethics Committee of Drum Tower Hospital. Animals were housed (6-7 per cage) with free access to food and water and were kept in temperature-controlled facilities at FIGURE 1 | Schematic presentation of experimental design. CHD, chow diet; HFD, high-fat diet; STZ, streptozotocin; Iso, isoflurane; Met, metformin; GS, 25% glucose solution; FC, fear conditioning test; WB, Western blot.

22-25
• C under a 12 h light/dark cycle. Diabetic mellitus type 2 (T2DM) mice model was prepared as we described previously (Fang et al., 2020). After 1 week of acclimatization, 7-week-old male C57BL/6J wild type (WT) mice from the Model Animal Research Center of Nanjing University were fed either chow diet or high-fat diet (HFD, 60% kcal fat, 20% carbohydrate). After fed with HFD for 2 weeks, inducible diabetic mellitus type 2 mice were fasted overnight and intraperitoneally injected with streptozotocin (STZ, Sigma-Aldrich, United States) freshly dissolved in 0.1 mM citrate buffer (pH 4.5) at a dose of 60 mg/kg body weight and tested for fasting 12 h tail vein blood glucose levels 1-week post-injection. Mice fed with chow diet were injected intraperitoneally with the citrate buffer vehicle. Mice with typical features of T2DM (polyphagia, polydipsia and polyuria) and fasting blood glucose levels more than 11.1 mmol/L were considered to be T2DM and were used in the experiment. Animals were maintained with their respective diets and body weight was measured every 3 days during the experiment. WT mice fed a standard chow diet were randomly divided into five subgroups receiving vehicle injection (Con), isoflurane anesthesia (Anes), metformin injection (Met), metformin pretreatment plus isoflurane anesthesia (Met + Anes), or metformin plus gavage 25% glucose solution prior to isoflurane (Glu + Met + Anes). Another group of T2DM model mice was randomly divided into a control group (DM), isoflurane anesthesia group (DM + Anes), metformin group (DM + Met), and metformin pretreatment plus isoflurane anesthesia group (DM + Met + Anes). Mice in Anes, Met + Anes, Glu + Met + Anes, DM + Anes, and DM + Met + Anes groups were exposed to isoflurane anesthesia in a chamber prefilled with 4% isoflurane (Lunan Better Pharmaceutical Co.) in 100% oxygen and then maintained with 1.3% isoflurane in 100% oxygen flowing at 2.5 L/min for 6 h. The respiratory activities of mice were monitored during anesthesia. To prevent the effect of hypothermia on tau protein phosphorylation, mice were placed on heating pads to maintain body temperature during anesthesia and lasted until full recovery. Mice in Met, Met + Anes, Glu + Met + Anes, DM + Met, and DM + Met + Anes groups were injected intraperitoneally with metformin (Sigma-Aldrich, D150959) 50 mg/kg body weight 1 h before anesthesia or at the corresponding time. Mice in the Glu + Met + Anes group were administered 25% glucose solution 0.1 ml (25%Glu 0.1 ml + Met + Anes) or 0.2 ml (25%Glu 0.2 ml + Met + Anes) by gavage before anesthesia (Figure 1).

Measurements of Blood Glucose
To eliminate interference from cage mates, mice were housed singly during tail vein blood sampling. Experienced researchers cautiously snip 1-2 mm of tissue from the tail tip distal with sharp scissors, and blood was obtained by gently massaging the tail. Squeezing the tail violently should be avoided because it might be painful for the mice and affect the quality of the sample. Stroking over the incision gently with sterile swabs dipped in saline reopened it and several blood samples could be collected in subsequent experiments. Blood glucose levels were tested by a hand-held glucometer (ACCU-CHEK Active, Roche, Switzerland).

Fear Conditioning Test
The fear conditioning test was conducted in 30 cm × 37 cm × 25 cm chambers equipped with a sound amplifier to produce auditory cues and a metal grid floor to deliver conditioning foot shocks. During the training phase, a mouse was placed in the chamber and allowed to explore freely for 3 min as habituation. After habitation, the sound (4,000 Hz, 100 dB) was presented for 30 s (conditioned stimulus, CS) and co-terminated with a foot shock (2 s, 0.8 mA) as an unconditional stimulus (US). Freezing behavior (a sign of conditioned fear) was scored for 1 min after foot shock. The chamber was then cleaned and descented with 75% ethanol between trials. Contextual fear conditioning was tested 24 h later in the same chamber by scoring freezing behavior (% of total time) for 3 min in the absence of sound or foot shock. A cued fear memory test was performed 2 h after the contextual fear conditioning test. Briefly, the sensory environment of the chamber (textures, odors, and colors) was modified and animals allowed 3 min of exploration as an accommodation period, followed by delivery of the CS (4,000 Hz, 100 db, 30 s). Freezing time (% of total time) was recorded automatically and the data were analyzed using Packwin 2.0 software.

Statistical Analysis
All data were parametric and summarized as mean ± standard deviation (SD). Results from cognitive behavioral tests, Western blotting, RT-PCR and light absorbance were analyzed by a oneway ANOVA test, followed by Bonferroni multiple comparison test. Changes in blood glucose levels over time were analyzed using multivariate analyses of variance for repeated measures followed by Bonferroni post hoc analysis. Statistical analysis was carried out using SPSS 25.0 software (IBM Corporation, Armonk, NY). Statistical significance referred to differences at the level of P < 0.05.
there were no significant effects on cued fear conditioning, suggesting that these effects are selective for hippocampal function ( Figure 2B).

Metformin Inhibits Hippocampal Tau Hyperphosphorylation and Increases NR2B Tyr1472 Phosphorylation in Isoflurane-Anesthetized Mice
Clinical and preclinical studies have found that impaired insulin signaling induces tau hyperphosphorylation and that insulin-sensitizing strategies improve cerebrospinal fluid biomarkers of tauopathy in patients with MCI/AD (Hölscher, 2014;Yarchoan et al., 2014;. As a cytoskeleton-associated protein, tau plays an important role in postsynaptic targeting of the Src family tyrosine kinase Fyn to NMDA receptors and in phosphorylating the NR2B subunit at Tyr1472 (pNR2B-Tyr1472), which is associated with NMDA receptor-dependent plasticity (Ittner et al., 2010). Conversely, hyperphosphorylated tau FIGURE 4 | Metformin alleviates tau hyperphosphorylation and pNR2B-Tyr1472 downregulation caused by long-term isoflurane anesthesia. Representative western blots (left panels) and densitometric analysis (right panels, n = 3 mice per group) showing the effects of isoflurane anesthesia (Anes group), metformin (Met group), and anesthesia plus metformin pretreatment (Met + Anes group) on expression levels of total tau (Tau) and phosphorylated tau (Ser396 and AT8) (A), as well as total NR2B (NR2B) and phosphorylated NR2B at Tyr1472 (pNR2B-Tyr1472) (B) in the hippocampus. β-Actin was used as a gel loading control. Data are expressed as mean ± SD and compared by one-way ANOVA followed by Bonferroni multiple comparisons tests. * * P < 0.01 compared to the Con group; && P < 0.01 compared to the Anes group.

DISCUSSION
Our present study demonstrated that prolonged (6 h) isoflurane anesthesia can induce hippocampal insulin resistance and impair hippocampus-dependent contextual fear memory in adult WT mice. Further, these deleterious effects were exacerbated in HDF/STZ-induced T2DM model mice. Intraperitoneal pretreatment with metformin (50 mg/kg body weight) attenuated isoflurane-induced hippocampal IR and contextual fear memory impairment in WT mice, but had a more blunted effect in T2DM model mice. Metformin also reduced hippocampal tau hyperphosphorylation and increased expression of hippocampal pNR2B-Tyr1472, which may have contributed to the improvement in hippocampus-dependent memory. Metformin also successfully reversed isofluraneinduced hyperglycemia, but elimination of glycemic control by high oral glucose did not influence the effects on IR and memory impairment, suggesting a novel independent mechanism for the therapeutic benefits of this antidiabetic drug.

Isoflurane and HFD/STZ Synergistically Induce Hippocampus-Dependent Cognitive Decline
As T2DM is one of the most common, costly, and disabling conditions in the industrialized world (Cho et al., 2018), we focused on the role of preoperative IR created by HFD/STZ treatment and postoperative IR caused by long-term isoflurane inhalation on POCD. The combination of HFD feeding and low-dose STZ treatment is a classic method for inducing a T2DM-like phenotype in mice (Srinivasan et al., 2005).
Indeed, our previous study demonstrated that HFD/STZ could induce marked peripheral and brain IR in mice (Fang et al., 2020). Brain insulin action is required for neuronal survival, synaptic plasticity, and cognitive function (van der Heide et al., 2006;Biessels and Reagan, 2015;Grillo et al., 2015). Both adult WT mice and T2DM mice demonstrated impairments in contextual fear conditioning after long-term isoflurane inhalation. However, isoflurane had no effects on cued fear conditioning. While the amygdala is critical for the expression of both cued and contextual conditioned fear responses, contextual fear memories are consolidated and maintained by the hippocampus (Chaaya et al., 2018). Thus, isoflurane and HFD/STZ treatment appear to selectively impair hippocampusdependent fear learning and memory. Compared to adult WT mice receiving anesthesia (Anes group), T2DM model mice (DM + Anes group) displayed significantly weaker contextdependent freezing responses on Day 1 and a very weak freezing response on Day 3 that did not differ after long-term isoflurane inhalation, suggesting that contextual fear learning and memory were more severely impaired in T2DM mice. This is consistent with clinical studies reporting that diabetes patients are more susceptible to POCD (Feinkohl et al., 2017). Thus, preoperative IR and postoperative IR have synergistic effects on POCD.

Metformin Attenuated Hippocampal IR and Tau Hyperphosphorylation and Increase Hippocampal pNR2B-Tyr1472 Expression
To examine potential mechanisms underlying the memory impairments induced by isoflurane and HFD/STZ treatment, we measured the effects of the first-line antidiabetic drug metformin on cognitive performance, IR, and phosphorylation of the NMDA receptor, which is essential for many forms of synaptic plasticity underlying hippocampus-dependent learning and memory (Rostas et al., 1996;Petrone et al., 2003). Metformin has been shown to cross the blood-brain barrier rapidly and effectively ameliorate IR (Giannarelli et al., 2003;Łabuzek et al., 2010). Increases in IRS1 and IRS2 serine phosphorylation, as well as decreases in downstream Akt and GSK3β phosphorylation are hallmark features of brain IR (Fang et al., 2020). Consistent with hippocampal IR, long-term isoflurane inhalation increased pIRS1-Ser639 and pIRS2-Ser731 expression levels and downregulated pIRS1-Tyr896, pAkt-Ser473, and pGSK3β-Ser9 expression levels in the hippocampus, while metformin reversed these responses. A similar pattern of changes was also observed in longterm isoflurane anesthetized T2DM mice. Concomitant with improved hippocampal IR, metformin pretreatment ameliorated the impairment in fear learning induced by long-term isoflurane inhalation in adult WT mice, but was less effective in T2DM mice. These results are consistent with numerous studies implicating hippocampal IR as a key mediator of cognitive dysfunction (Reviewed in Biessels and Reagan, 2015). Clinical and preclinical studies have also revealed that impaired insulin signaling induces tau hyperphosphorylation and that insulin-sensitizing strategies improve cerebrospinal fluid biomarkers of disease in patients with mild cognitive impairment or AD (Hölscher, 2014;Yarchoan et al., 2014;. Consistent with these clinical findings, isoflurane upregulated the expression of p-tau in hippocampus, while metformin decreased the expression of hippocampal p-tau. As a cytoskeleton-associated protein, tau plays an important role in the postsynaptic targeting of the Scr kinase Fyn to NMDA receptors, where it phosphorylates the NR2B subunit at Tyr-1472 and promotes NMDAR-dependent synaptic plasticity (Rostas et al., 1996;Petrone et al., 2003;Ittner et al., 2010). Moreover, hyperphosphorylated tau has been shown to reduce Fyn binding (Reynolds et al., 2008). Metformin not only attenuated IR, but also inhibited the phosphorylation of tau protein and prevented isofluraneinduced downregulation of pNR2B-Tyr-1472, providing a feasible mechanism for the improvement in contextual fear memory following metformin treatment. Therefore, metformin pretreatment may effectively alleviate hippocampal IR, prevent tau hyperphosphorylation, enhance the phosphorylation of Tyr-1472 in the NR2B, and reverse cognitive impairment caused by long-term isoflurane inhalation.

Metformin Mitigated Cognitive Decline Through a Mechanism Independent of Glycemic Control
Previous studies have shown that hyperglycemia contributes to cognitive decline in diabetes patients (Cox et al., 2005). Higher glycated hemoglobin, an index of more severe and prolonged hyperglycemia, is associated with a greater magnitude of cognitive impairment, and even acute hyperglycemia can impair cognitive performance in persons with type 1 or type 2 diabetes mellitus (Sommerfield et al., 2004;Gao et al., 2015;Šuput Omladič et al., 2020). Intraoperative hyperglycemia also increases the risk of POCD (Puskas et al., 2007;Schricker et al., 2014). Our previous animal studies and related clinical studies have shown that isoflurane inhalation significantly increases blood glucose (Behdad et al., 2014;Fang et al., 2020). In addition to relieving insulin resistance, metformin reduces blood sugar by suppressing hepatic glucose production, increasing glucose uptake into muscle, and promoting the transport of blood glucose into stool (Giannarelli et al., 2003;Morita et al., 2020). Our present studies did find that intraperitoneal injection of metformin (50 mg/kg body weight) not only decreased blood glucose in adult WT and T2DM mice, but also dramatically reversed the hyperglycemic effect of isoflurane during anesthesia. Surprisingly, however, glycemic control did not contribute to the improvement in hippocampus-dependent fear memory under the current experimental conditions, as gavage treatment with 25% glucose solution elevated blood glucose even under metformin treatment, but did not influence metforminassociated improvements in insulin signaling pathways and fear learning following isoflurane anesthesia. The mechanism by which acute hyperglycemia causes cognitive decline remains unclear. Whether hyperglycemia impair cognitive performance through insulin signaling need further study.

Pretreatment of Metformin Had No Therapeutic Effects on HFD/STZ-Induced T2DM
Our present study found that metformin had no effects on the activation of hippocampal insulin signaling and intraperitoneal injection of metformin also had no effects on cognitive performance of T2DM mice. However, preclinical studies have shown that metformin treatment improved cognitive deficits in db/db mice and chronic HFD-fed mice (Li et al., 2012;Pintana et al., 2012;Chen et al., 2019). Clinical studies also demonstrated that long-term usage can lower the risk of cognitive impairment in older adults with diabetes (Ng et al., 2014). Different research models and drug treatment times may explain this discrepancy. Ramos-Rodriguez and colleagues found that the central pathologies and cognitive impairments differed among diabetes models (Ramos-Rodriguez et al., 2013). Db/db mice may be more amenable to metformin treatment that C57BL/6J mice used in this study and longer-term HFD treatment (12 weeks) was used for T2DM model animals in previous studies (Li et al., 2012;Pintana et al., 2012;Chen et al., 2019). In previous studies, metformin administration time has varied from 21 days or 8 weeks (Li et al., 2012;Pintana et al., 2012;Chen et al., 2019). Clinical studies reported that short-term metformin treatment (36 weeks) had no effect on cognitive performance in diabetes patients, while longer-term treatment (more than 6 years) was beneficial (Abbatecola et al., 2010;Ng et al., 2014). Whether longterm metformin treatment has therapeutic effects on HFD/STZinduced T2DM mice requires further study.

CONCLUSION
Hippocampal IR, tau hyperphosphorylation, and downregulation of pNR2B-Tyr1472 contribute to prolonged isoflurane-induced cognitive dysfunction. Brief metformin treatment can mitigate these effects through a mechanism independent of glycemic control. However, a single intraperitoneal injection of metformin had less therapeutic effects on cognitive decline induced by isoflurane in T2DM mice. A single-dose pretreatment of metformin had no effects on the changes of hippocampal IR and cognitive decline induced by HFD/STZ. Future studies are needed to investigate whether long-term metformin treatment can also prevent HFD/STZ -induced hippocampal IR and cognitive decline.

DATA AVAILABILITY STATEMENT
The raw data supporting the conclusions of this article will be made available by the authors, without undue reservation.

ETHICS STATEMENT
The studies involving human participants were reviewed and approved by the Medical Ethics Committee of the Affiliated Drum Tower Hospital of Medical Department of Nanjing University. The patients/participants provided their written informed consent to participate in this study. The animal study was reviewed and approved by Laboratory Animal Ethics Committee of Drum Tower Hospital. Written informed consent was obtained from the individual(s) for the publication of any potentially identifiable images or data included in this article.

AUTHOR CONTRIBUTIONS
XGu, TX, and ZM conceived the original idea. FX, XF, and SL performed the animal experiments. FX, XGo, and XF carried out the clinical study. LP, YQ, and XZ generated and analyzed data and wrote the manuscript with input from all authors. All authors read and approved the final manuscript. (C) Freezing time to cue by adult WT mice receiving vehicle (Anes), metformin (Met + Anes), or 25% glucose 0.1 ml plus metformin (Glu + Met + Anes) prior to isoflurane anesthesia (n = 7-8). Data are expressed as mean ± SD and analyzed with one-way ANOVA test followed by Bonferroni multiple comparison test.