Adiponectin pathway regulates cerebral metabolic dysfunction and neuroinflammation via the AdipoR1/PI3K/Akt axis in Perioperative Neurocognitive Disorder

Abstract

Backgroud

Perioperative neurocognitive disorder (PND) is a prevalent and serious complication in elderly surgical patients, with limited effective therapeutic options available. While our prior research has demonstrated the neuroprotective potential of the adiponectin pathway in PND, the underlying mechanisms remain to be fully elucidated.

Methods

In a prospective cohort study, we collected serum, cerebrospinal fluid (CSF), and sociodemographic data from 41 elderly hip fracture patients (29 normal and 12 PND patients). Further, twelve-month-old male Sprague-Dawley rats were divided into sham, PND (splenectomy), and PND + Adiporon (APN, 50 mg/kg/day intragastrically) group. Lactate, pyruvate, TNF-α and IL-1β levels in CSF and hippocampus were measured. Additionally, a PND + APN + LY294002 (a PI3K inhibitor, 25 mg/kg/day intraperitoneally) group was established to explore the underlying mechanisms further. Cognitive function was assessed using the Morris Water Maze (MWM) test. Glucose transport (Glut) 1, glycolysis (HK2, PFKFB3 and PKM2), energy production (ATP and Na⁺/K⁺-ATPase), microglia-mediated neuroinflammation (Iba1, TNF-α, IL-1β) and synaptic protein (PSD95, SYP and SYN I) were assessed in hippocampus.

Results

PND elderly patients exhibited lower serum adiponectin levels, which correlated with higher lactate/pyruvate ratio (Pearson’s r correlation: -0.4513; p = 0.0031) and higher TNF-α level (Pearson’s r correlation: -0.4311; p = 0.0049) in CSF. In PND rats, APN reduced lactate, lactate/pyruvate ratio, TNF-α, and IL-1β in brain. Mechanistically, APN activated AdipoR1-dependent PI3K/Akt signaling, enhanced Glut1 membrane localization, HK2 activity, and Na⁺/K⁺-ATPase activity. APN also inhibited microglia overactivation and neuroinflammation. Activation of the adiponectin pathway improved cognitive performance in the MWM test.

Conclusion

The adiponectin pathway regulates cerebral metabolic dysfunction and neuroinflammation via the AdipoR1/PI3K/Akt axis, which serves as a potential therapeutic target for improving perioperative cognitive outcomes in elderly patients.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12877-025-06367-w.

Introduction

Perioperative neurocognitive disorder (PND) presents a critical clinical challenge in elderly patients undergoing surgery or anesthesia, resulting in various cognitive impairments, such as memory loss, attention deficits, and executive function disorders. These conditions not only complicate postoperative recovery but also diminish long-term quality of life [1–3]. Conventional management strategies include improving perioperative sleep quality, administering dexmedetomidine or haloperidol, measuring the depth of anesthesia and so on [2, 4]. However, current therapeutic strategies remain insufficient, underscoring the need for mechanism-driven therapies.

In recent years, accumulating evidence has highlighted cerebral metabolic dysfunction and neuroinflammation as crucial factors in the pathogenesis of cognitive deficits [2, 5]. Impaired glucose and oxygen supply result in energy deficits, which further exacerbate neuroinflammation [6]. Neuroinflammation, marked by microglial activation and the release of pro-inflammatory cytokines such as interleukin-1β (IL-1β) and tumor necrosis factor-α (TNF-α), impairs mitochondrial function and energy supply [7]. These two interconnected processes create a vicious cycle, intensifying neuronal damage and cognitive decline.

Adiporon (APN), an orally active adiponectin receptor agonist, is involved in glucose, lipid, and energy metabolism. APN could permeate the blood-brain barrier (BBB) and modulate neuronal function [8]. AdipoR1 and AdipoR2, the receptors of the adiponectin pathway, are widely distributed in the central nervous system (CNS), primarily in neurons and blood vessels [8, 9]. Clain et al. discovered that AdipoRs are expressed in all brain regions, with AdipoR1 showing higher expression levels [8]. Recently, attention has focused on APN’s impact in the context of PND. Our researches have demonstrated a decrease in circulating adiponectin levels in PND animals [10] and human patients [11, 12]. Notably, we found that activation of the adiponectin pathway could alleviate disease progression [10]. These findings provide a basis for targeting adiponectin pathway as a therapeutic strategy for PND. However, the underlying mechanism of its neuroprotective effect remains unknown.

In this context, the present study was designed to explore the impact and the mechanism of adiponectin pathway on cerebral glycolysis and neuroinflammation after surgical trauma and anesthesia, providing new insights into improving perioperative cognitive outcomes in elderly patients.

Materials and methods

Participants

PND occurs with high prevalence in elderly patients who receive elective hip replacement surgery [13]. We enrolled those hip fracture patients aged 65 to 85 years old with an American Society of Anesthesiologists grade I to II classification, and without a history of alcohol or drug dependence, use of sedatives or antidepressants in the past two years, renal dysfunction, active liver disease, abnormal coagulation function shown in preoperative biochemical examination, or a history of central nervous system or mental illness. Patients with severe visual or hearing impairments, and those who could not communicate effectively or did not undergo anesthesia according to the research plan, requested to withdraw from the study midway were excluded.

This study was approved by the Dongguan People’s Hospital Ethics Committee (KYKT2022-07) and registered in the Chinese Clinical Trial Registry (ChiCTR2100051734) on October 1, 2021. The written informed consent was obtained from all participants. All patients received a standardized anesthesia procedure. A subarachnoid block anesthesia was performed at the L3/4 interspace using 1.5 ml of 0.5% bupivacaine. Based on the surgical duration, supplemental doses of 5 ml of 0.25% ropivacaine were administered epidurally every 1.5 h as needed. Norepinephrine was adjusted according to the baseline blood pressure (± 20%) intraoperatively. Epidural morphine 1 mg was administered for postoperative analgesia.

Blood samples were collected 5 min before spinal anesthesia. CSF was collected in propylene tubes at the onset of spinal anesthesia. Following collection, the samples were centrifuged, aliquoted, and stored at −80 °C. Permission of the Mini-Mental State Examination (MMSE) has been granted from Parinc (PAR). MMSE was administered at 10:00 a.m. on postoperative day (POD) 1, 3 and 7 by the same researcher, who had undergone training under the supervision of experienced neuropsychologists. A decline in the MMSE score to 24 was regarded as indicative of cognitive decline [14].

Animals

Twelve-month-old male Sprague-Dawley rats were obtained from the Model Animal Research Institute at Nanjing University (Nanjing, China). The study protocol involving animals was reviewed and approved by the Institutional Animal Care and Use Committee of Dongguan People’s Hospital, and it adhered to the relevant ARRIVE guidelines.

Rats were housed in a controlled environment with a temperature range of 24 ± 2 °C and a relative humidity of 55 ± 5%. They were maintained on a reversed 12-h light/dark cycle and given unrestricted access to standard rodent chow. These rats began training in the Morris Water Maze (MWM) test to evaluate whether they developed PND. One animal in the PND and PND + APN + LY294002 groups respectively died on POD 7, and their data were excluded from the analysis. The remaining rats were euthanized for test. Two independent replications of each experiment were performed. No outliers were removed prior to analysis.

Surgical procedure

The rats received isoflurane anesthesia, which is induced with 3.0% isoflurane and maintained with 1.5% isoflurane [15]. A blanket warmed the animals to maintain a body temperature of 37 °C. During surgery, we made a small incision in the abdomen, then followed by spleen isolation and removal. After splenectomy, the skin was sutured, and the animals were placed in warmed cages. 50 mg/kg cefoperazone (5%) was intraperitoneally injected after surgery [16]. Compound lidocaine cream (2.5% lidocaine and 2.5% prilocaine) was applied to the wound for pain relief, and the next application was 3 h after the operation and 10 min before the start of the behavior tests.

Treatment

Our prior research indicates that there is no significant difference between APN-treated and untreated sham animals [10]. APN (Proteintech, 924416-43-3) was dissolved in 1% DMSO and sterile saline, then administered via intragastric (i.g.) gavage at a dosage of 50 mg/kg/d [17] from 7 days prior to surgery/anesthesia until sample collection. LY294002 (PI3K inhibitor, Proteintech, 154447-36-6) in sterile saline was given intraperitoneally at 25 mg/kg/d [18], following the timeline from 7 days before to 7 days after surgery/anesthesia.

The Morris water maze

The pool, measuring 1.2 m in diameter, was filled with water to a depth of 50 centimeters, maintained at 23 ± 1 °C. Based on four equidistant marks (East, South, West, North), their projections on the water surface and bucket bottom divided the water area and bucket into four equal quadrants. The platform was positioned 1 cm above the water surface in clear water within quadrant II. Five days prior to surgery, all rats underwent training with a visible platform. If a rat failed to locate the platform within 60 s, it was gently guided to the platform and allowed to remain there for 30 s. Starting from the fourth day before surgery, the rats were trained in navigation trials. Each rat underwent four trials per day for four consecutive days. The water was made opaque by adding non-toxic white dye to obscure the platform’s location. Rats were released into the maze facing the pool wall to standardize the starting orientation. A trial was deemed successful if the rat reached and remained on the platform for at least 3 s. After each trial, rats were allowed to remain on the platform for 30 s to reinforce the spatial memory. Subsequently, on POD 3, 5, and 7, the hidden platform was removed, and the rats were tested for spatial exploration. During these probe trials, the swimming speed, the number of quadrant crossings, and the escape latency were recorded within a 60 s period.

Sample collection

Rats were sacrificed via intraperitoneal injection of ketamine (120 mg/kg) and xylazine (10 mg/kg). On the POD 1 and 7, six rats in each group were used for ELISA, colorimetric analysis, biochemical analysis, western blot and quantitative polymerase chain reaction.

To collect the CSF, the skin was incised from the back to the head. The overlying connective tissue was removed and the skull was exposed. Using an insulin syringe, freehand puncture was performed carefully to prevent brain stem damage and blood contamination. The whole hippocampus hemisphere was rapidly isolated and quick-frozen in liquid nitrogen and stored at − 80 °C until analysis.

Meanwhile, three rats in each group were used for immunofluorescence analysis. Transcardial perfusion was conducted initially with precooled saline, followed by 4% paraformaldehyde for fixation.

ELISA

Serum and CSF were rapidly frozen on powdered dry ice, and stored at − 80 °C until processing. The IL-1β (Elabscience, E-EL-R0012c) and TNF-α (Elabscience, E-EL-R2856c) levels were determined. Analysis was carried out with ELISA kits following the manufacturer’s instructions.

Colorimetric analysis

After collection, CSF samples were immediately stored at −80 °C and assayed within 24 h. Lactate (Elabscience, E-BC-K044-S) and pyruvate (Elabscience, E-BC-K130-S) levels were measured by colorimetric analysis (CMA600 Microdialysis Analyzer) following the manufacturer’s instructions.

Biochemical determination

The frozen hippocampus tissues were homogenized in nine-fold volumes of cold double-distilled water, and the homogenate was centrifuged at 3500×g and 4 °C for 10 min. The supernatants were collected for determination of protein concentration. Then the homogenate was boiled for 10 min, mixed well, recentrifuged at 3500×g for 10 min, and the supernatant was taken for testing. Adenosine triphosphate (ATP) and ATPase levels were determined using commercial biochemical kits following the manufacturer’s instructions. (Jiancheng Bioengineering Institute, Nanjing, China).

Immunofluorescence

Each block was sectioned at 20 μm, heated at 60 °C for 30 min, deparaffinized in xylene (5 min × 3 times), and dehydrated in 100%, 95%, and 70% ethanol (3 times each). Sections were then treated with 3% hydrogen peroxide in methanol, blocked with 5% BSA at room temperature for 1 h, frozen at 4 °C, and subsequently stained.

Slides were first incubated in 5% BSA at 37 °C for 1.5 h, followed by overnight incubation with anti-Iba1 antibody (1:100, Abiowell, AWA11292) in 0.5% BSA at 4 °C. After rinsing in PBS (3 times), slides were incubated with secondary antibody at 37 °C for 1 h and then mounted with glass coverslips. Images were captured using an Olympus BX53 upright microscope. Sections were manually analyzed at 400× magnification by 2 researchers blinded to the treatment groups.

Western blotting

Western blotting was conducted following a previously established protocol [19]. Briefly, the blots were blocked with 5% skim milk at room temperature for 1.5 h, then incubated overnight at 4 °C with primary antibodies, followed by incubation with HRP-conjugated secondary antibodies at room temperature for 1.5 h. Protein bands were visualized using a Tanon 5200 imaging system, and densitometric analysis was carried out with Quantity One Software. Details of antibodies are shown in suppl Table 1.

Quantitative polymerase chain reaction

Total RNA was extracted from the hippocampus using TRIzol reagent following the manufacturer’s instructions (Invitrogen, Carlsbad, CA). cDNA synthesis was carried out with the M-MLV First-Strand Synthesis Kit (CWBIO, Beijing, China). For mitochondrial DNA (mtDNA) detection, total DNA from the hippocampus was extracted using a DNA Extraction Kit (CWBIO, Beijing, China). qPCR was conducted using a SYBR Green PCR Kit (CWBIO, Beijing, China) on an ABI Step One Plus system (Applied Biosystems, Foster City, CA, USA). Primers, synthesized by Invitrogen (Shanghai, China), are listed in Supplementary Table 2. GAPDH was the internal reference gene for gene expression normalization, while β-globin served as the nuclear reference for mtDNA normalization. Results were analyzed using the 2 − ΔΔCT method.

Statistical analysis

All statistical analyses were performed using GraphPad Prism 5.0 and IBM SPSS version 25. Descriptive statistics were used to summarize the continuous variables, presented as means with standard error of means (SEMs) or medians with quartiles (Qs). Categorical variables were presented as frequencies and percentages. Normality of distribution was assessed via the Kolmogorov-Smirnov test. Variables such as the lactate (LA)/pyruvate (PA) ratios in both groups, operating time and blood loss in the PND group exhibited non-normal distribution and were analyzed using Mann-Whitney U tests. Binary logistic regression analyses were carried out for the associations of PND with serum adiponectin, CSF TNF-α, and the ratio of LA/PA with no multicollinearity between the independent variables (VIF < 1.5). The remaining data were evaluated by t - tests or ANOVA. Post - hoc comparisons following ANOVA were conducted using Tukey’s test with Bonferroni correction, and adjusted p-values are reported for all multiple comparisons. A two-tailed p-value of less than 0.05 was regarded as statistically significant.

Result

Cognitive disorder in elderly patients was related to adiponectin deficiency mediated- glucose metabolic disorder and neuroinflammation

In clinical, PND is a common complication in elderly patients with hip fractures [13]. To explore the underlying mechanism of cognitive disorder in the context of human biology, 47 of these patients were selected and 41 patients were ultimately enrolled (Fig. 1). They were assessed by the MMSE and divided into normal group (29 patients) and PND group (12 patients). And we compared those clinical characteristics (Table 1), CSF and blood biomarkers (Fig. 2) among these 41 elderly patients.

Fig. 1.

Flowchart of selection for hip fracture patient’s cohort

Table 1.

Baseline of the study population

Characteristic	Normal (n = 29)	PND (n = 12)	p valued
Median age (y)	72.55 ± 0.9078	75 ± 1.532	0.2941
Male, n (%)	10 (34.48%)	4 (30.77%)	0.620
Educational level(y)	6.483 ± 0.3496	6.462 ± 0.4615	0.8071
Preoperative MMSE scores	25.86 ± 0.3048	25.67 ± 0.3761	0.8030
Hospital stay(Days)	11.83 ± 0.6842	16.69 ± 2.489	0.0162*
Bleeding (ml) M (P25,P75)	270.7(125, 300)	257.7(100, 300)	0.8851
Operating time(min) M (P25,P75)	153.1(115, 185)	158.5(105, 170)	0.2941

Fig. 2.

The relationship between serum adiponectin, CSF ratio of LA/PA, and TNF-α in patients with cognitive disorder. A Schematic illustration of sample collection from the cohort of patients. The levels of serum adiponectin (B), the LA/PA ratio (C), and TNF-α (D) in CSF. Correlation analysis of serum adiponectin with CSF LA/PA (E) and TNF-α (F). ^*p < 0.05, ^**p < 0.01, ^***p < 0.001 vs. the normal patients

First, there were no significant differences in age, gender, education level, preoperative MMSE score, operation duration, and bleeding between the PND group and normal group. However, PND significantly prolonged hospital stays (Table 1). Meanwhile, in PND patients, the level of serum adiponectin was markedly decreased (Fig. 2B), the LA/PA ratio (Fig. 2C) and TNF-α level (Fig. 2D) in CSF were increased. Binary logistic regression analyses showed that high serum adiponectin (OR = 0.305, p = 0.020, 95%CI [0.112–0.829]) was independent protective factor for PND, while high LA/PA ratio (OR = 12.131, p = 0.006, 95%CI [2.071–19.075]) and high TNF-α level (OR = 1.925, p = 0.029, 95%CI [1.071–3.461]) in CSF were independent risk factors for PND (Table 2). Additionally, lower serum adiponectin level was significantly correlated with higher CSF LA/PA ratio (Pearson’s correlation: −0.4513; p = 0.0031, Fig. 2E) and higher TNF-α level (Pearson’s correlation: −0.4311; p = 0.0049, Fig. 2F).The receiver operating characteristic (ROC) curve for the PND group is presented in Table 3. The area under the curve (AUC) for adiponectin in predicting PND was 0.7284 (95% CI [0.5633, 0.8936]), indicating that adiponectin has preliminary predictive potential.

Table 2.

Logistic regression analysis of serum adiponectin, CSF TNF-α, and CSF LA/PA

Biomarkers	B	SE	Wald	p value	Odd Ratio	95%CI
Serum Adiponectin	−1.186	0.510	5.422	0.020*	0.305	0.112–0.829
CSF TNF-α	0.655	0.299	4.790	0.029*	1.925	1.071–3.461
CSF LA/PA	2.496	0.902	7.656	0.006**	12.131	2.071–19.075

Table 3.

ROC analysis for serum adiponectin in PND

	AUC	95%CI	Sensitivity	Specificity	Youden index	Optimum critical value	p value
Serum Adiponectin	0.7284	0.5633-0.8936	91.70%	48%	0.39	6.7619	0.02277

APN alleviated cerebral metabolic disorder and neuroinflammation in PND rats

Then, we explored whether administered APN could regulate glucose metabolism and neuroinflammation in rats. We found that surgical trauma significantly led to acute cerebral metabolic disorder (Fig. 3 A, B) and neuroinflammation (Fig. 3 C, D) on POD 1. And the PND rats treated with APN exhibited lower LA/PA ratio (Fig. 3A), lactate (Fig. 3B), TNF-α (Fig. 3 C) and IL-1β (Fig. 3D) in CSF, indicating that peripheral application of APN could effectively regulate cerebral glucose metabolism and neuroinflammation early. And neuroinflammation could persist up to POD 7, with the levels of hippocampal TNF-α and IL-1β in PND rats remained higher than those in the sham group (suppl Fig. 1).

Fig. 3.

APN regulated acute cerebral metabolic disorder and neuroinflammation in PND rats. The ratio of LA/PA (A), the levels of lactate (B), TNF-α (C) and IL-1β (D) in CSF. n = 6. ^***p < 0.001 vs. the sham group after Bonferroni correction; ^&&p < 0.01 vs. the PND group after Bonferroni correction

APN activated the AdipoR1/PI3K/Akt signaling pathway in PND rats

Next, we explored the underlying mechanism of APN regulation. We first detected the expression of both its receptors in hippocampus. Results showed that AdipoR1 (Fig. 4A, B), expression was reduced in the PND group compared to the sham group, and these changes were corrected under APN treatment. Recent data have revealed that the PI3K/Akt signaling is the downstream target of the adiponectin signaling pathway [20, 21]. Then, we investigated the phosphorylation of PI3K/Akt. The PI3K/Akt pathway was significantly inhibited in PND (Fig. 4A, C, D). APN activated the AdipoR1/PI3K/Akt pathway, an effect that was prevented by LY294002, a PI3K inhibitor (Fig. 4A, C, D).

Fig. 4.

APN activated the AdipoR1/PI3K/Akt pathway in PND rats. A Representative hippocampal AdipoR1, p-PI3K, PI3K, p-Akt, Akt and β-actin western blot images. Protein expression of hippocampal AdipoR1 (B), p-PI3K (C) and p-Akt (D). n = 6. ^*p < 0.05, ^**p < 0.01 vs. the sham group after Bonferroni correction; ^&p < 0.05, ^&&p < 0.01 vs. the PND group after Bonferroni correction; ^#p < 0.05, ^##p < 0.01 vs. the PND + APN group after Bonferroni correction

APN enhanced the translocation of glucose transporter 1, the glycolytic enzymes, and ATPase activity in PND rats via the PI3K pathway

To elucidate APN-mediated metabolic restoration via PI3K signaling, we systematically examined key glycolytic components including glucose transporter Glut1, rate-limiting enzymes, and ATP production. We found decreased Glut1 membrane location in PND group (Fig. 5A, B). Regarding the expression of glycolytic enzymes, decreased glycolytic enzyme hexokinase 2 (HK2) and increased phosphofructokinase-2/fructose-2,6-bisphosphatase 3 (PFKFB3) were pronounced in PND (Fig. 5C-E), whereas no significant differences were observed in the expression of pyruvate kinase M2 (PKM2) (Fig. 5C, F). APN increased membrane location of Glut1 and the expression of HK2 in PND, but had no significant effect on PFKFB3 (Fig. 5C, E). Concurrently, the reduced Na⁺/K⁺-ATPase activity observed in PND could also be elevated by APN (Fig. 5G), while no significant difference was found in ATP levels (Fig. 5H). The aforementioned effects could be prevented by LY294002 (Fig. 5A-D, G).

Fig. 5.

APN regulated hippocampal metabolic function in PND via the PI3K pathway. A Representative hippocampal Glut1 and ATP1A1 western blot images. B Protein expression of Glut1. C Representative hippocampal HK2, PFKFB3, PKM2 and β-actin western blot images. Protein expression of HK2 (D), PFKFB3 (E) and PKM2 (F). Activity of Na⁺/K⁺- ATPase (G) and level of ATP (H) in hippocampus. n = 6. ^*p < 0.05, ^**p < 0.01 vs. the sham group after Bonferroni correction; ^&p < 0.05 vs. the PND group; ^#p < 0.05 vs. the PND + APN group after Bonferroni correction

APN inhibited microglia-mediated neuroinflammation in PND rats via the PI3K pathway

Meanwhile, we investigated the effects of APN on microglial activation via the PI3K pathway. In the PND group, microglia in the hippocampal region were significantly activated, with upregulated expression of microglial marker, ionized calcium binding adaptor molecule 1 (Iba1) (Fig. 6A, B). Immunofluorescence assays also showed that the cell bodies of microglia became enlarged and their processes were retracted (Fig. 6C). APN could downregulate the expression of Iba1 (Fig. 6A, B). The over-activated state of microglia was ameliorated, with their processes becoming slender and cell bodies shrinking (Fig. 6C). Also, APN inhibited hippocampal TNF-α and IL-1β (Fig. 6D, E). Inhibition of the PI3K signaling pathway could reverse the APN-mediated alleviation of neuroinflammation.

Fig. 6.

APN inhibited microglia-mediated neuroinflammation in PND via the PI3K pathway. A Representative hippocampal Iba1 western blot images. B Protein expression of Iba1. C Representative hippocampal microglia immunofluorescence staining images (n = 3, scale bar, 100 μm). The gene expression of TNF-α (D) and IL-1β (E) in the hippocampus. ^**p < 0.01 vs. the sham group after Bonferroni correction. n = 6. ^&p < 0.05, ^&&p < 0.01 vs. the PND group after Bonferroni correction; ^#p < 0.05 vs. the PND + APN group after Bonferroni correction

APN enhanced postsynaptic function postsurgery in PND rats via the PI3K pathway

Neuronal synaptic function could directly influence cognitive and behavioral performance. We examined the changes in synaptic-related proteins in PND and found that the expression of the postsynaptic skeleton protein PSD95 was significantly decreased (Fig. 7A, B), whereas the expression levels of synaptic vesicles proteins, SYP (Fig. 7C) and SYN1 (Fig. 7D) remained unchanged. APN could elevate PSD95 expression, and antagonizing PI3K could partially abrogate the effects of APN (Fig. 7A, B).

Fig. 7.

APN augmented postsynaptic proteins in PND via the PI3K pathway. A Representative hippocampal PSD95, SYP, SYN I and β-actin western blot images. Protein expression of hippocampal PSD95 (B), SYP (C) and SYN1 (D). ^*p < 0.05 vs. the sham group after Bonferroni correction; ^&p < 0.05 vs. the PND group after Bonferroni correction; ^#p < 0.05 vs. the PND + APN group after Bonferroni correction

APN alleviated surgery-associated cognitive disorders via the PI3K pathway

Then, we utilized the MWM test to assess the effects of APN on cognitive deficits via PI3K pathway. Rats in the PND group had a longer latency and fewer platform crossings compared to the sham group on POD 3 (Fig. 8A-D). These results indicated that postsurgical rats developed cognitive deficits from POD 3 onwards. However, although APN could increase the number of platform crossings in PND animals on POD3, it did not shorten their latency period until the fifth day. The long-term effect of APN could reduce the latency to find the target platform, increase the number of platform crossings, and thereby enhance cognitive performance (Fig. 8C, D). However, LY294002 prevented the beneficial effects of APN (Figs. 8).

Fig. 8.

APN ameliorated surgery-associated cognitive disorders. Representative traces of rat movement trajectories (A), swimming speed (B), the number of platform crossings (C) and escape latency (D) in the MWM test. ^*p < 0.05, ^**p < 0.01 vs. the sham group after Bonferroni correction; ^&p < 0.05 vs. the PND group after Bonferroni correction; ^#p < 0.05 vs. the PND + APN group after Bonferroni correction

Discussion

Surgical trauma or anesthesia can alter cerebral energy metabolism, leading to chemotaxis in an energy-deficient state and aggravating neuroinflammation and neuronal damage. These are the main pathogeneses of cognitive disorders following surgical trauma/anesthesia. The present study demonstrates that APN ameliorates perioperative neurocognitive disorders through modulation of cerebral glucose metabolism and microglial-mediated neuroinflammation via the AdipoR1/PI3K/Akt signaling pathway. Our findings highlight novel therapeutic potential of targeting adiponectin signaling in PND.

Under normal physiological conditions, glucose is primarily transported into the brain via Glut1 and then phosphorylated by HK2 to generate glucose-6-phosphate (G6P) and fructose-6-phosphate (F6P). Subsequently, PFKFB3 catalyzes the interconversion between F6P and fructose-2, 6-bisphosphate (F2,6BP), thereby enhancing the activity of phosphofructokinase-1. PKM2 facilitates the generation of pyruvate, which is converted to acetyl-CoA by pyruvate dehydrogenase to produce ATP. However, in PND, this cascade is disrupted. We observed lactate accumulation, reduced Na⁺/K⁺-ATPase activity, consistent with a shift toward anaerobic metabolism and mitochondrial dysfunction [15, 22].

In this study, we confirmed that APN restored hippocampal glucose metabolism by enhancing Glut1 membrane translocation and specifically augmenting HK2 activity, thereby rescuing Na⁺/K⁺-ATPase activity. Previous studies have shown that surgery-induced reductions in Glut1 significantly contribute to postoperative cognitive deficits in rats by impairing cerebral glucose transport [16]. Meanwhile, HK2, the predominant mitochondria -related isozyme, is selectively expressed in microglia. Genetic ablation of HK2 reduced microglial glycolytic flux and energy production. Moreover, HK2 elevation is prominent in immune-challenged or disease-associated microglia [23]. It has been demonstrated that enhancing HK2 expression can improve cognitive deficits by increasing glucose metabolism in Alzheimer’s disease [24]. Here, we confirmed that peripheral administration of APN selectively restored early glycolysis, as evidenced by its specific upregulation of HK2 activity without significant effects on PFKFB3 or PKM2. This underscores the pivotal role of HK2 as a metabolic gatekeeper in microglia and identifies it as a key target through which APN rectifies bioenergetic deficits in PND.

Generally, the trends of changes in diseases involving HK2 and PFKFB3 are consistent. However, we found that HK2 and PFKFB3 exhibited opposite trends in PND, which is different from other neurological disorders [24, 25]. This paradoxical response likely stems from compensatory metabolic feedback and inflammation-driven reprogramming. Specifically, HK2 downregulation impairs glycolytic initiation by reducing G6P and F6P production [26], potentially triggering compensatory PFKFB3 upregulation to salvage glycolytic flux. Concurrently, PFKFB3 elevation aligns with its established pro-inflammation role [27, 28], where PND-associated neuroinflammation may directly drive pathological overexpression, creating a maladaptive feedback loop that exacerbates metabolic dysfunction [29]. Future studies will investigate this mechanism through in vitro assessment of PFKFB3 dynamics following G6P, F6P, or LPS challenge.

APN attenuated microglial hyperactivation and suppressed pro-inflammatory cytokines, extending the previous reports of its anti-inflammatory properties [30]. Critically, our finding demonstrates APN’s ability to penetrate the blood-brain barrier and directly modulate neuroimmune responses. By disrupting the self-perpetuating cycle of neuroinflammation and energy deficits, APN emerges as a promising therapeutic strategy for PND.

Mechanistically, PI3K/Akt signaling inhibition abolished APN’s protective effects, positioning this pathway as a central regulator. This aligns with emerging paradigms of metabolic-inflammatory crosstalk in neurological disorders [26, 31, 32]. Phosphorylated Akt not only enhances Glut1 trafficking [33] but also inhibits NF-κB nuclear translocation [34], thereby coupling metabolic rescue to inflammasome repression. This dual regulatory capacity explains why LY294002 prevented both ATPase restoration and cytokine reduction.

Moreover, APN rescued postsynaptic PSD95 expression. The selective restoration of PSD95, implied that adiponectin signaling preferentially stabilizes postsynaptic architecture, possibly via Akt-dependent mTOR activation—a pathway implicated in dendritic spine formation [35]. Meanwhile, APN did not significantly improve MWM performance until POD 5. It is possible that prolonged APN treatment may fully exert its neuroprotective effects.

Finally, there are some limitations in our study that need to be noted. First, this prospective cohort study was conducted at a single center with a relatively small sample size. Meanwhile, MMSE has limitations in differentiating between specific subtypes of PND, which limits the generalizability of the results. Second, cerebral energy metabolism is a complex process involving glucose, lipid metabolism and so on. In this study, we clarified the impact of APN on glucose metabolism in the whole brain. Future research could use single-cell metabolomics to further explore the effects on individual cell types. Thirdly, estrogen could influences the adiponectin signaling pathway [36]. The inclusion of female animals and human cohorts is essential for future investigation. Lastly, different experimental groups with various APN doses should be designed to study its dose-response relationship.

Conclusion

In conclusion, APN could restore metabolic defect and suppress neuroinflammation via AdipoR1/PI3K/Akt signaling. Adiponectin pathway offers a novel approach to perioperative neuroprotection and holds potential for improving cognitive outcomes in elderly surgical patients.

Supplementary Information

Below is the link to the electronic supplementary material.

Author contributions

H.X. conceived and designed the experiments. Z.Z. performed ELISA, qPCR and colorimetric analysis. Z.Z. and C.H. performed western blotting and immunofluorescence assays. Z.Z., C.H., B.L. and C.Z. performed the surgery, all administrations and behavioral testing. L.G. and H.X. collected clinical samples. B.S. and H.C. analyzed the data. H.X. and Z.Z. wrote the paper. All authors read and approved the final manuscript.

Funding

This work was supported by Basic and Applied Basic Research Foundation of Guangdong Province (grant no. 2023A1515012456), Guangdong Medical Research Foundation (grant no. A2024387) and Guangdong Provincial Medical Association Anesthesiology Branch Research Foundation (grant no. GDSA202202004).

Data availability

The data that support the findings of this study are available on request to the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

Declarations

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.