Esketamine Attenuates Postoperative Neurocognitive Disorder in Aged Mice: Associations with Bioenergetic Remodeling in Hippocampal CD11b-Enriched Cell Fractions and Glycolysis-Related Signaling Authors: Di Liu, Guangchun Li, Hui Huang, Yong Yang, Fei Jiang, Yue Zhang Published: 2026 ABSTRACT AbstractBackgroundPostoperative neurocognitive disorder (PND) is a clinically important complication in older surgical patients. Microglial inflammatory activation is frequently accompanied by altered cellular metabolism, but whether esketamine-associated neuroprotection is accompanied by bioenergetic changes in hippocampal CD11b-enriched cell fractions remains incompletely understood.MethodsA tibial fracture model was used to induce PND in aged male mice. The public hippocampal transcriptomic dataset GSE165798 was analyzed as an exploratory, hypothesis-generating screen. Independent animal cohorts were evaluated using behavioral testing, whole-hippocampal qRT-PCR and biochemical assays, immunofluorescence, transmission electron microscopy, and Seahorse extracellular flux analysis of hippocampal CD11b-enriched cell fractions. Early hippocampal IL-1β was measured at 24 h in a tissue-analysis cohort. In a separate contemporaneous five-group pharmacological cohort, behavioral outcomes were assessed before hippocampal IL-1β and TNF-α measurement on postoperative day 14. Systemic 2-DG was used as a non-cell-specific pharmacological probe of associations between glycolytic inhibition and selected outcomes.ResultsExploratory transcriptomic analysis suggested coordinated glycolysis-related and inflammatory changes in the PND hippocampus. In independent cohorts, esketamine improved spatial learning and memory, attenuated the early 24-h hippocampal IL-1β elevation, preserved synaptic ultrastructure, and reduced glycolysis-related molecular and biochemical changes measured in whole hippocampal tissue. In hippocampal CD11b-enriched cell fractions collected at 24 h, esketamine was associated with lower glycolytic activity and improved mitochondrial respiratory parameters. In the separate pharmacological cohort, hippocampal IL-1β and TNF-α were measured on day 14 after behavioral testing. Systemic 2-DG was associated with selected behavioral and day-14 cytokine differences relative to untreated PND mice; a contemporaneous Control + 2-DG group was included. In direct comparisons of the esketamine and 2-DG groups, the mean differences were −0.40s for time in the target quadrant (95% CI, −4.35 to 3.55; Tukey-adjusted P = 0.9984), 4.71 pg/mg protein for IL-1β (95% CI, −8.99 to 18.41; Tukey-adjusted P = 0.8588), and 2.38 pg/mg protein for TNF-α (95% CI, −12.69 to 17.45; Tukey-adjusted P = 0.9908). These nonsignificant comparisons neither establish therapeutic equivalence nor demonstrate that glycolytic inhibition mediates the effects of esketamine.ConclusionEsketamine attenuated surgery-induced cognitive impairment, the early 24-h hippocampal IL-1β elevation, and synaptic ultrastructural damage in aged mice. In a separate pharmacological cohort assessed on day 14, esketamine and systemic 2-DG were each associated with lower hippocampal IL-1β and TNF-α than untreated PND. These findings were accompanied, in separate experimental cohorts, by improved bioenergetic profiles in hippocampal CD11b-enriched cell fractions and reduced glycolysis-related markers in whole hippocampal tissue. The data support associations among esketamine treatment, metabolic-inflammatory remodeling, and neuroprotection, but do not establish a microglia-specific HK2/PKM2-dependent mechanism, a causal role for glycolysis, or mechanistic equivalence between esketamine and 2-DG.Keywords: esketamine, postoperative neurocognitive disorder, microglia, bioenergetics, glycolysis, hippocampus Introduction Postoperative neurocognitive disorder (PND) encompasses cognitive decline identified during the perioperative period and is particularly relevant to older surgical patients, in whom it may impede functional recovery and quality of life.1,2 Its pathogenesis is multifactorial, but perioperative immune activation and neuroinflammation are consistently implicated. Surgical trauma can initiate peripheral inflammatory signaling and activate hippocampal microglia, with downstream release of mediators such as IL-1β that may disrupt synaptic plasticity and contribute to cognitive impairment.3 Microglia are not only immune effector cells but also metabolically adaptable cells whose functions depend on the balance between glycolysis and mitochondrial oxidative phosphorylation.4–7 Inflammatory stimulation can increase glycolytic flux and rapidly support cytokine production, whereas persistent mitochondrial dysfunction and loss of metabolic flexibility may sustain maladaptive inflammatory states.8,9 However, prior in vivo evidence does not support a uniformly detrimental interpretation of microglial glycolysis. In surgical-trauma and other neuroinflammatory models, increased glycolysis has been associated with proinflammatory microglial responses and cognitive or tissue injury, and suppressing excessive glycolytic activation has produced benefit.10–12 By contrast, metabolic adaptation can also preserve microglial phagocytic or reparative functions and help maintain neuronal network activity during moderate or early inflammation; prolonged disease may instead produce metabolic exhaustion and impaired microglial function.8,13 These apparently divergent observations indicate that the consequences of glycolytic reprogramming depend on disease model, inflammatory intensity, cell state, and time point. HK2 and PKM2 are therefore biologically plausible candidate regulators, but their involvement cannot be assumed to be uniformly pathogenic, microglia-specific, or causally required in the postoperative brain. Esketamine, the S-enantiomer of ketamine, is used clinically as an anesthetic and analgesic and has shown anti-inflammatory and neuroprotective effects in experimental and perioperative studies.14–20 Proposed mechanisms include attenuation of microglial inflammatory activation, preservation of synaptic function, modulation of autophagy and cell-death pathways, and signaling effects beyond N-methyl-D-aspartate receptor antagonism.14,16–19 Nevertheless, existing studies differ in disease model, dose, timing, and metabolic readout, and most infer microglial metabolic regulation from tissue-level markers rather than direct extracellular-flux measurements in freshly isolated cells. A traumatic brain injury study suggested that esketamine can influence glycolysis-associated immune responses,17 but it remains unknown whether this observation extends to the aged postoperative brain or whether behavioral protection is accompanied by restoration of bioenergetic flexibility in hippocampal CD11b-enriched cell fractions. This distinction is important for translation: glucose metabolism is essential to microglia, neurons, astrocytes, endothelial cells, and peripheral immune cells, so nonselective glycolytic inhibition may disrupt physiological metabolic homeostasis.13,21 Accordingly, the present study investigated whether esketamine-associated improvement in surgery-induced cognitive impairment is accompanied by bioenergetic changes in hippocampal CD11b-enriched cell fractions and by hippocampal glycolysis-related signaling. We first used the small public hippocampal transcriptomic dataset GSE165798 as an exploratory, hypothesis-generating screen. We then evaluated the resulting metabolic-inflammatory hypothesis in independent cohorts of aged mice using behavioral testing, hippocampal molecular and biochemical assays, immunofluorescence, transmission electron microscopy, and extracellular flux analysis of hippocampal CD11b-enriched cell fractions. Finally, systemic 2-deoxy-D-glucose was used as a non-cell-specific pharmacological probe to examine whether glycolytic inhibition was associated with selected behavioral and inflammatory outcomes. This design was intended to integrate complementary evidence while distinguishing exploratory transcriptomic findings, whole-hippocampal molecular changes, bioenergetic measurements in CD11b-enriched cell fractions, and associative pharmacological observations that do not establish pathway mediation. Materials and Methods Animals and Ethical Approval Aged male C57BL/6 mice (18 months old, weighing 30–40 g) were obtained from the Laboratory Animal Center of Chongqing Medical University (Chongqing, China). Animals were housed in a specific pathogen-free facility under controlled conditions (23 ± 2°C; relative humidity, 55 ± 5%) with a 12-h light/dark cycle and ad libitum access to standard chow and water. All mice were acclimatized for at least 1 week before experimental procedures. Male mice were used in this initial study to reduce sex-related hormonal and immunometabolic variability within the available sample size; this design does not address potential sex-dependent treatment effects. All experimental protocols were approved by the Ethical Committee of Chongqing Medical University (Approval No. 2024–49). Experiments were conducted in accordance with the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All efforts were made to minimize animal suffering and to use the minimum number of animals compatible with independent biological replication. At the prespecified collection endpoints, mice were deeply anesthetized with 7–8% sevoflurane in oxygen until loss of the pedal withdrawal reflex, followed by cervical dislocation as a secondary physical method. Death was confirmed by the absence of spontaneous respiration and heartbeat before tissue collection. Establishment of the Perioperative Neurocognitive Disorder Model and Control Condition The PND model was established using tibial fracture surgery as previously described, with minor modifications. Mice were anesthetized with 7–8% sevoflurane for induction and maintained with 3–4% sevoflurane through a facemask. Body temperature was maintained at 37–38°C using a heating pad throughout the procedure, and physiological status was monitored during anesthesia. After shaving and disinfecting the left hind limb, a median incision was made over the left tibia. A 0.38-mm intramedullary pin was inserted into the tibial canal, followed by periosteal stripping and osteotomy using surgical scissors. The incision was sutured and treated with 1.0% ropivacaine (0.1 mL) for local analgesia. Mice in the Control group underwent sevoflurane anesthesia, perioperative handling, temperature maintenance, and local analgesia but did not receive skin incision, periosteal stripping, osteotomy, or tibial fracture. This anesthesia-only control was selected to account for anesthetic exposure and perioperative handling while preserving the contrast with the complete surgical-injury model. A separate sham-operated group was not included; therefore, the individual contributions of incision, soft-tissue manipulation, periosteal injury, and fracture could not be distinguished. Drugs and Reagents Esketamine hydrochloride was purchased from Hengrui Medicine Co., Ltd. (Jiangsu, China). Esketamine was administered intraperitoneally at 5 mg/kg immediately after induction of anesthesia during tibial fracture surgery. This esketamine dose was selected on the basis of previous studies reporting neuroprotective and anti-inflammatory effects in rodent models of perioperative neurocognitive dysfunction and brain injury.14,19 Mice in the PND + normal saline group received an equivalent volume of normal saline intraperitoneally immediately after anesthesia induction, at the same administration time point as the esketamine-treated group. The glycolytic inhibitor 2-deoxy-D-glucose (2-DG) was obtained from Sigma-Aldrich (St. Louis, MO, USA). Primary antibodies used for immunofluorescence included mouse IgG1 chimeric monoclonal anti-Iba1 [EPR16589] (1:1000; Abcam, Cat. No. ab283319, Cambridge, UK) and rabbit monoclonal anti-GLUT1 [EPR3915] (1:200; Abcam, Cat. No. ab115730, Cambridge, UK). Alexa Fluor 594-conjugated anti-mouse IgG and Alexa Fluor 488-conjugated anti-rabbit IgG secondary antibodies were obtained from Thermo Fisher Scientific (Waltham, MA, USA). The Mouse IL-1β/IL-1F2 Quantikine ELISA Kit (Cat. No. MLB00C) and Mouse TNF-α Quantikine ELISA Kit (Cat. No. MTA00B) were obtained from R&D Systems (Minneapolis, MN, USA). L-lactate, pyruvate, pyruvate kinase, and lactate dehydrogenase were measured using Abcam assay kits (L-Lactate Assay Kit, Cat. No. ab65330; Pyruvate Assay Kit, Cat. No. ab65342; Pyruvate Kinase Assay Kit, Cat. No. ab83432; and LDH Assay Kit, Cat. No. ab102526; Abcam, Cambridge, UK). Experimental Design, Randomization, Blinding, and Sample Allocation Mice were assigned to experimental groups using a computer-generated randomization sequence. Investigators responsible for behavioral testing, imaging acquisition, outcome quantification, biochemical assays, and statistical analysis were blinded to group allocation. Group codes were retained until data collection and prespecified analyses were completed. The study comprised three experimental cohorts. In the first cohort, 10 mice per group were used for behavioral testing in the Control, PND, PND + esketamine, and PND + normal saline groups. After behavioral testing, five mice per group were allocated to transmission electron microscopy and the remaining five mice per group were allocated to immunofluorescence analysis. Separate animals were used for early tissue collection at 24 h after surgery; six independent mice per group contributed whole-hippocampal samples for qRT-PCR, IL-1β ELISA, lactate and pyruvate measurements, and pyruvate kinase and lactate dehydrogenase activity assays. Tissue aliquots from the same animal were used across these molecular and biochemical assays, and each mouse was counted once as one biological replicate. Thus, the first cohort comprised 16 mice per group (10 mice for behavioral testing and subsequent TEM or immunofluorescence, plus 6 independent mice for 24-h tissue assays), corresponding to 64 mice in total. The second cohort was used for acute preparation of hippocampal CD11b-enriched cell fractions and Seahorse extracellular flux analysis. Hippocampi from six non-overlapping mice within the same treatment group were pooled to generate one independent biological sample. Six independent biological pools were analyzed per group, corresponding to 36 mice per group. Each biological pool, rather than an individual contributing mouse or a technical replicate well, was treated as the unit of statistical analysis. The third cohort evaluated systemic 2-DG in a contemporaneous five-group design. Mice were assigned to Control, PND, PND + esketamine, PND + 2-DG, or Control + 2-DG groups, with 10 mice per group undergoing the same randomized, blinded behavioral-testing schedule. Before behavioral testing, eight of the 10 mice in each group were predesignated by computer-generated endpoint allocation for hippocampal cytokine analysis after the Morris water maze; selection was not based on behavioral performance or other observed outcomes. Hippocampal IL-1β and TNF-α were measured in these eight mice per group on postoperative day 14. All five groups were enrolled and studied contemporaneously under the same husbandry, randomization, blinding, behavioral-assessment, and tissue-collection procedures. Thus, the third cohort comprised 50 mice. Across the three cohorts, 258 mice were used in total (64 in the first cohort, 144 in the second cohort, and 50 in the third cohort). Predefined exclusion criteria included death unrelated to the planned endpoint, inability to complete the prespecified surgical model, severe postoperative complications that precluded scheduled assessment, and technical failure during sample processing or assay acquisition. No animals or samples were excluded for technical, clinical, or outcome-related reasons after randomization. All animals were included in their assigned endpoint analyses; the lower n for post-behavioral cytokine measurements reflected endpoint-specific sample allocation rather than post-randomization exclusion. Behavioral Assessment Spatial learning and memory were assessed using the Morris water maze. The apparatus consisted of a circular black pool (120 cm in diameter and 50 cm high) filled with water maintained at 24 ± 1°C and rendered opaque with non-toxic titanium dioxide. A hidden platform (10 cm in diameter) was submerged 1 cm below the water surface in the target quadrant. Mice underwent acquisition training on postoperative days 8–12, with four trials per day. During each trial, mice were allowed 60s to locate the hidden platform. A mouse that failed to locate the platform within 60s was gently guided to it and allowed to remain there for 15s. Escape latency was recorded as an index of spatial learning. On postoperative day 13, the platform was removed for a 60-s probe trial. Time spent in the target quadrant, number of platform crossings, swimming speed, and representative swimming paths were recorded using the ANY-maze video-tracking system (Stoelting Co., Wood Dale, IL, USA). Swimming speed was analyzed to assess potential motor-performance confounding. The open field test was performed on postoperative day 7 to assess locomotor activity and anxiety-like behavior. Total distance traveled and time spent in the center zone were recorded using an automated tracking system. Behavioral testing and analysis were performed by investigators blinded to treatment allocation. Transmission Electron Microscopy Hippocampal CA1 tissue blocks (1–2 mm3) were dissected, fixed in 2.5% glutaraldehyde and 1% osmium tetroxide, dehydrated through a graded ethanol series, and embedded in epoxy resin. Ultrathin sections were stained with uranyl acetate and lead citrate and examined using a JEOL JEM-1400 Plus transmission electron microscope (Japan Electron Optics, Japan). Synaptic numerical density and postsynaptic density thickness were quantified by investigators blinded to group allocation. Matched images were acquired at consistent magnification, and scale bars were generated from the original microscope calibration metadata. Immunofluorescence Brain sections (20 μm) were prepared to evaluate Iba1-positive microglia and their spatial relationship with GLUT1-positive microvasculature. After blocking with 5% normal goat serum and 0.3% Triton X-100, sections were incubated overnight at 4°C with mouse anti-Iba1 and rabbit anti-GLUT1 antibodies. Sections were then washed with phosphate-buffered saline and incubated for 1 h at room temperature with Alexa Fluor 594-conjugated anti-mouse IgG and Alexa Fluor 488-conjugated anti-rabbit IgG secondary antibodies, respectively. Nuclei were counterstained with DAPI. Images were acquired using an Olympus FV3000 confocal laser-scanning microscope under identical acquisition settings across treatment groups. Scale bars were calibrated from microscope metadata and applied consistently to matched images. Representative images were selected by investigators blinded to group allocation and were qualitatively inspected to describe the spatial distribution of Iba1-positive cells relative to GLUT1-positive microvasculature; no formal fluorescence-intensity or perivascular-clustering quantification was performed. This descriptive assay evaluated Iba1/GLUT1 spatial association only; Iba1/HK2 and Iba1/PKM2 co-localization analyses were not performed. Quantitative Real-Time PCR Total RNA was extracted from whole hippocampal tissue using TRIzol reagent. RNA concentration and purity were measured using a NanoDrop 2000 spectrophotometer (Thermo Fisher Scientific). RNA was reverse-transcribed into cDNA using MuLV reverse transcriptase and oligo(dT) primers, and quantitative real-time PCR was performed using SYBR Green Premix. Relative mRNA expression was calculated using the 2^−ΔΔCt method and normalized to GAPDH after confirming stable GAPDH expression across groups. The analyzed glycolysis-related transcripts included Hk2, Ldha, Pdk1, and Pkm2. Primer sequences are provided in Table S1. These measurements were performed in whole hippocampal homogenates, not in acutely isolated microglia. HK2 or PKM2 expression was not directly measured in isolated microglia, and no Iba1/HK2 or Iba1/PKM2 co-localization assay was conducted. Accordingly, the molecular measurements were interpreted as hippocampal tissue-level changes rather than evidence of microglia-specific HK2 or PKM2 regulation. Measurement of Lactate and Pyruvate Levels Lactate and pyruvate concentrations were measured in whole hippocampal tissue homogenates using the L-Lactate Assay Kit (Abcam, Cat. No. ab65330) and Pyruvate Assay Kit (Abcam, Cat. No. ab65342), respectively, according to the manufacturer’s protocols. Absorbance was measured using a microplate reader, concentrations were calculated from standard curves, and values were normalized to total protein concentration. Assessment of Pyruvate Kinase and Lactate Dehydrogenase Activities Pyruvate kinase and lactate dehydrogenase activities were measured in whole hippocampal tissue homogenates using the Pyruvate Kinase Assay Kit (Abcam, Cat. No. ab83432) and LDH Assay Kit (Abcam, Cat. No. ab102526), respectively. Enzyme activities were calculated from kinetic absorbance changes according to the manufacturers’ instructions and normalized to total protein concentration. ELISA Measurement of IL-1β and TNF-α Hippocampal tissues were homogenized in ice-cold lysis buffer and centrifuged to collect the supernatant. IL-1β and TNF-α were measured using the Mouse IL-1β/IL-1F2 Quantikine ELISA Kit (R&D Systems, Cat. No. MLB00C) and Mouse TNF-α Quantikine ELISA Kit (R&D Systems, Cat. No. MTA00B), respectively. Optical density was measured using a microplate reader, concentrations were calculated from standard curves, and values were normalized to total protein and expressed as pg/mg protein. The first tissue-analysis cohort provided an early 24-h IL-1β measurement (n = 6 per group). In the separate pharmacological cohort, both IL-1β and TNF-α were measured on postoperative day 14 after completion of behavioral testing (n = 8 per group). Measurements from these distinct cohorts and time points were analyzed and interpreted separately. The two mediators were not interpreted as a comprehensive neuroinflammatory panel. CD11b-Positive Cell Enrichment and Seahorse Extracellular Flux Analysis Hippocampal CD11b-positive cells were enriched 24 h after surgery. This time point was selected to characterize the early postoperative inflammatory–metabolic response before later recovery-related changes became dominant, consistent with previous PND studies examining early hippocampal responses.10,22 The 24-h analysis was not intended to establish the temporal peak of glycolytic activation in the enriched cell fraction. For each independent biological sample, hippocampi from six non-overlapping mice in the same treatment group were pooled. Tissue was enzymatically dissociated using the Adult Brain Dissociation Kit, mouse and rat (Miltenyi Biotec, Cat. No. 130–107-677, Bergisch Gladbach, Germany). CD11b-positive cells were enriched using CD11b MicroBeads (Cat. No. 130–093-634) and LS Columns (Cat. No. 130–042-401) with magnetic-activated cell sorting. Samples were maintained at approximately 4°C during dissection and purification whenever feasible, and all isolation procedures were completed within 2 h. The resulting CD11b-enriched cell fraction was used for extracellular flux analysis. An independent flow-cytometric or immunocytochemical quantification of post-enrichment purity was not performed; therefore, the preparation is described as CD11b-enriched rather than assigned an unverified numerical purity. The CD11b-enriched cell fraction was seeded at 5.0 × 10^4 cells per well into Agilent Seahorse XF HS miniplates and analyzed using an Agilent Seahorse XF HS Mini Extracellular Flux Analyzer (Agilent Technologies, Santa Clara, CA, USA). The XFp Glycolysis Stress Test Kit (Cat. No. 103017–100) and XFp Cell Mito Stress Test Kit (Cat. No. 103010–100) were used according to the manufacturer’s instructions. The sensor cartridge was hydrated overnight in Seahorse XF Calibrant at 37°C in a non-CO2 incubator. Before each assay, the culture medium was replaced with 180 μL of prewarmed assay medium per well, and cells were equilibrated at 37°C in a non-CO2 incubator for 45–60 min. Wells A and H were used as cell-free background correction wells. Each biological pool was analyzed in three technical replicate wells, and replicate values were averaged before statistical analysis. For the Glycolysis Stress Test, cells were incubated in glucose- and pyruvate-free Seahorse XF DMEM medium (pH 7.4) supplemented with 2 mM L-glutamine. ECAR was recorded at baseline and after sequential injection of glucose (10 mM final concentration), oligomycin (1.0 μM), and 2-DG (50 mM). For the Mito Stress Test, Seahorse XF DMEM medium (pH 7.4) was supplemented with 10 mM glucose, 2 mM L-glutamine, and 1 mM sodium pyruvate. OCR was recorded at baseline and after sequential injection of oligomycin (1.5 μM), FCCP (1.0 μM; within the manufacturer-recommended titration range), and rotenone/antimycin A (0.5 μM each). Three measurement cycles were acquired at baseline and after each injection, with each cycle consisting of 3 min of mixing followed by 3 min of measurement. Glycolytic parameters included basal glycolysis, glycolytic capacity, and glycolytic reserve; mitochondrial parameters included basal respiration, ATP-linked respiration, and maximal respiration. ECAR and OCR values were normalized to total cellular protein per well using a BCA assay. Technical replicate wells from the same biological pool were averaged before statistical analysis. Pharmacological Intervention with 2-DG A separate contemporaneous cohort was used to examine associations between systemic glycolytic inhibition and selected behavioral and hippocampal cytokine outcomes. Mice were randomized to Control, PND, PND + esketamine, PND + 2-DG, or Control + 2-DG groups. The PND + 2-DG and Control + 2-DG groups received 2-DG (500 mg/kg, intraperitoneally) once daily on study days 0–2. Control + 2-DG mice underwent the same anesthesia, perioperative handling, temperature maintenance, and local analgesia as the Control group but did not undergo surgical incision, periosteal manipulation, osteotomy, or fracture. The PND + esketamine group received esketamine as described in Drugs and Reagents. All five groups were enrolled contemporaneously and underwent the same computer-generated randomization, investigator blinding, monitoring, behavioral-testing schedule, and day-14 tissue-collection workflow. The Control + 2-DG group was included as a non-surgical comparator for the measured behavioral and hippocampal cytokine endpoints. Its inclusion improves assessment of 2-DG-associated effects under the Control condition but does not establish metabolic safety, cell specificity, or absence of unmeasured systemic effects. Accordingly, the 2-DG experiment was interpreted as an associative, non-cell-specific pharmacological observation rather than proof that glycolytic activation caused the PND phenotype, mediated the effects of esketamine, or represented an equivalent therapeutic mechanism. Bioinformatics and Transcriptomic Analysis The publicly available hippocampal transcriptomic dataset GSE165798 was obtained from the Gene Expression Omnibus. Computational analyses and visualizations were performed using R version 4.3.2. Raw expression data were log2-transformed and quantile-normalized. Differential expression analysis was conducted using limma version 3.58.1. Differentially expressed genes were defined using a Benjamini–Hochberg adjusted P value <0.05 and an absolute log2 fold change >1. Volcano plots and heatmaps were generated using ggplot2 version 3.4.4 and pheatmap version 1.0.12. Gene Ontology and Kyoto Encyclopedia of Genes and Genomes enrichment analyses were performed using clusterProfiler version 4.10.1. Pathways related to glycolytic metabolism, inflammation, and synaptic function were prioritized for hypothesis generation. GSE165798 contains three Sham and three PND samples according to the original study design. Therefore, this analysis was treated solely as an exploratory, hypothesis-generating screen and not as independent confirmatory evidence or cell-type-specific validation. Statistical Analysis Statistical analyses were performed using GraphPad Prism 10.1.2 (GraphPad Software, San Diego, CA, USA). Data distributions and variance assumptions were assessed before parametric testing. Normally distributed continuous data are presented as mean ± SEM. Comparisons among multiple groups were performed using one-way ANOVA followed by Tukey’s post hoc test. Morris water maze acquisition data were analyzed using two-way repeated-measures ANOVA with treatment group and training day as factors, followed by multiplicity-adjusted post hoc comparisons. For the contemporaneous five-group 2-DG cohort, behavioral and day-14 hippocampal cytokine outcomes were analyzed separately using one-way ANOVA followed by Tukey-adjusted multiple comparisons. Comparisons of primary interest included Control versus Control + 2-DG, PND versus PND + esketamine, PND versus PND + 2-DG, and PND + esketamine versus PND + 2-DG. Because variance or residual-distribution diagnostics indicated departures from parametric assumptions for selected outcomes, sensitivity analyses were also performed. Welch one-way ANOVA with Games–Howell comparisons was used for target-quadrant time and IL-1β, and the Kruskal–Wallis test with Holm-adjusted pairwise Mann–Whitney tests was used for TNF-α. Because the study was not designed as an equivalence or non-inferiority experiment, nonsignificant comparisons were not interpreted as evidence of equivalent efficacy. For the whole-hippocampal molecular and biochemical assays, six independent mice per group were analyzed, and each mouse was treated as one biological observation. For Seahorse assays, six independent biological pools per group were analyzed, and each pool was treated as one observation. Technical replicate wells from the same pool were averaged before group-level analysis to avoid pseudoreplication. The early 24-h IL-1β assay used six independent mice per group, whereas the separate day-14 pharmacological cytokine assays used eight predesignated mice per group. The exact number of animals, independent biological samples or pools, and technical replicates was reported for each experiment. The public GSE165798 dataset was analyzed separately as an exploratory external dataset because its sample size was determined by the original study. All tests were two-sided, and P < 0.05 was considered statistically significant. Results Exploratory Transcriptomic Screening Suggests a Pro-Glycolytic and Inflammatory Signature in PND Hippocampus To identify candidate molecular pathways associated with PND, we first analyzed the publicly available hippocampal transcriptomic dataset GSE165798 following the bioinformatic workflow shown in Figure 1A. Because this dataset contains n = 3 Sham and n = 3 PND samples according to the original study design, the analysis was interpreted as a hypothesis-generating screen rather than as stand-alone confirmatory evidence. Exploratory transcriptomic and cell-reference analyses identify glycolysis-related and inflammatory signatures associated with PND. (A) Analysis workflow for the public hippocampal transcriptomic dataset GSE165798, including data acquisition, preprocessing, differential-expression analysis, targeted gene profiling, and pathway-level screening. (B) Volcano plot of the exploratory differential-expression analysis; representative glycolysis-related genes, including Hk2, Pkm, and Slc2a1, are highlighted. (C) Heatmap of selected glycolysis-related and inflammatory genes across Sham and PND samples. (D) Pathway-cluster analysis grouping selected genes into glycolysis-, inflammation-, and stress-related modules. (E) Supportive cell-type reference analysis of Hk2 expression using an independent, publicly available adult mouse brain single-cell RNA-sequencing resource. GSE165798 contained three Sham and three PND samples according to the original study design. These public-data analyses were exploratory and hypothesis-generating; they were not treated as cell-type-specific or independent confirmatory evidence. Volcano plot analysis suggested upregulation of several glycolysis-associated genes, including Hk2, Pkm, and Slc2a1, in the hippocampus of PND mice (Figure 1B). Heatmap analysis further showed differential expression of selected glycolytic and inflammatory genes across Sham and PND samples (Figure 1C). Pathway-level analysis indicated enrichment of glycolysis-related and inflammatory response modules (Figure 1D). These exploratory findings provided a rationale for examining metabolic-inflammatory coupling in PND. To obtain supportive cell-type reference information, we queried publicly available adult mouse brain single-cell RNA-sequencing resources. This analysis suggested that Hk2 expression is relatively enriched in microglial populations compared with several other major CNS cell types under the reference conditions examined (Figure 1E). Because these single-cell data were not generated from the same PND model, the results were considered supportive rather than definitive evidence for microglial involvement. Together, the public transcriptomic and reference single-cell analyses guided subsequent validation experiments in independent animal cohorts. Esketamine Alleviates Surgery-Induced Spatial Learning and Memory Impairment To determine whether esketamine protects against surgery-induced cognitive impairment, behavioral assessments were performed after tibial fracture surgery (Figure 2A). Open field testing on postoperative day 7 showed no significant differences in total distance traveled or time spent in the center zone among groups (Figure 2B–D; P > 0.05), indicating that locomotor activity and anxiety-like behavior were unlikely to confound subsequent cognitive testing. Esketamine attenuates surgery-induced spatial learning and memory deficits without altering locomotor activity. (A) Experimental timeline showing tibial fracture surgery and the single intraoperative esketamine dose on postoperative day 0, open-field testing on day 7, Morris water maze acquisition on days 8–12, the probe trial on day 13, and tissue collection thereafter. (B) Representative open-field movement tracks. (C) Total distance traveled. (D) Time spent in the center zone. (E) Representative swimming trajectories during the Morris water maze probe trial. (F) Mean swimming speed. (G) Escape latency during acquisition training. (H) Time spent in the target quadrant during the probe trial. Data are presented as mean ± SEM; n = 10 independent mice per group. One-way ANOVA was used for single-time-point outcomes, and two-way repeated-measures ANOVA was used for acquisition data, followed by Tukey-adjusted post hoc comparisons as appropriate. Asterisks indicate comparisons with the Control group, whereas number signs indicate comparisons with the untreated PND group. ****P < 0.0001 versus Control; ####P < 0.0001 versus untreated PND; ns, not significant. Representative swimming trajectories during the Morris water maze probe trial are shown in Figure 2E. Swimming speed did not differ significantly among groups (Figure 2F), suggesting that subsequent differences in task performance were not attributable to altered motor ability. During Morris water maze acquisition training, PND mice showed longer escape latencies than Control mice, whereas esketamine treatment shortened escape latencies relative to the untreated PND group (Figure 2G). In the probe trial on postoperative day 13, esketamine-treated mice spent more time in the target quadrant than untreated PND mice (Figure 2H), indicating improved spatial learning and memory. Esketamine Attenuates Hippocampal IL-1β Elevation and Glycolysis-Related Alterations The experimental timeline shows tibial fracture surgery on day 0, early hippocampal collection at 24 h for qRT-PCR, IL-1β ELISA, and biochemical assays, and immunofluorescence collection after behavioral testing (Figure 3A). Tibial fracture surgery increased hippocampal IL-1β levels in the PND group, and this increase was attenuated by esketamine (Figure 3B). IL-1β was selected as a representative inflammatory mediator because of its established role in synaptic plasticity disruption and relevance to PND models.3,10,22 qRT-PCR analysis showed that PND increased the expression of Hk2, Ldha, Pdk1, and Pkm2 in whole hippocampal tissue compared with the Control group (Figure 3C–F), whereas esketamine attenuated these increases. Whole-hippocampal lactate and pyruvate concentrations and lactate dehydrogenase and pyruvate kinase activities were also increased after surgery and attenuated by esketamine (Figure 3G–J). These tissue-level molecular and biochemical findings were analyzed in six independent mice per group and were not interpreted as microglia-specific HK2 or PKM2 measurements. Esketamine attenuates whole-hippocampal glycolysis-related changes and IL-1β elevation. (A) Experimental timeline indicating tibial fracture surgery on day 0, early hippocampal collection at 24 h for qRT-PCR, IL-1β ELISA, and biochemical assays, and later collection after behavioral testing for immunofluorescence. (B) Hippocampal IL-1β concentration measured by ELISA and normalized to total protein. (C–F) Whole-hippocampal Hk2, Ldha, Pdk1, and Pkm2 mRNA expression measured by qRT-PCR. (G–J) Whole-hippocampal lactate concentration, lactate dehydrogenase (LDH) activity, pyruvate concentration, and pyruvate kinase (PK) activity. (K) Representative CA1 immunofluorescence images showing Iba1-positive cells (red), GLUT1-positive microvasculature (green), and DAPI-stained nuclei (blue). Molecular and biochemical assays used n = 6 independent mice per group; immunofluorescence used n = 5 independent mice per group. Data are presented as mean ± SEM. Asterisks indicate comparisons with the Control group: *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Number signs indicate comparisons with the untreated PND group: #P < 0.05, ##P < 0.01, and ###P < 0.001. Hk2 and Pkm2 were measured in whole hippocampal tissue and should not be interpreted as microglia-specific measurements. Qualitative inspection of representative immunofluorescence images suggested greater spatial proximity of Iba1-positive cells to GLUT1-positive microvasculature in PND mice and a less prominent pattern in the PND + esketamine group (Figure 3K). Because this observation was descriptive and was not supported by formal image quantification, it was not interpreted as evidence of a treatment effect on perivascular microglial clustering. Esketamine Preserves Synaptic Ultrastructure and Is Associated with Improved Bioenergetic Profiles in Hippocampal CD11b-Enriched Cell Fractions Transmission electron microscopy showed that esketamine preserved hippocampal synaptic ultrastructure, as reflected by higher synaptic numerical density and postsynaptic density thickness than in untreated PND mice (Figure 4A–C). To examine cellular bioenergetics separately from the whole-hippocampal molecular measurements, hippocampal CD11b-positive cells were enriched 24 h after surgery and analyzed using the Agilent Seahorse XF HS Mini platform. In the Glycolysis Stress Test, the CD11b-enriched cell fraction from PND mice showed higher ECAR, basal glycolysis, and glycolytic capacity together with lower glycolytic reserve than the corresponding fraction from Control mice (Figure 4D and E). The PND + esketamine group showed lower glycolytic activity and partial recovery of glycolytic reserve relative to untreated PND mice. No significant difference was observed between the PND and PND + normal saline groups. Esketamine preserves synaptic ultrastructure and is associated with improved bioenergetic profiles in hippocampal CD11b-enriched cell fractions from PND mice. (A) Representative transmission electron microscopy images of hippocampal synapses; scale bars are indicated within the images. (B) Synaptic numerical density. (C) Postsynaptic density thickness. (D) Extracellular acidification rate (ECAR; left) during the Glycolysis Stress Test after sequential injection of glucose, oligomycin, and 2-DG, and oxygen consumption rate (OCR; right) during the Mito Stress Test after sequential injection of oligomycin, FCCP, and rotenone/antimycin A. (E) ECAR-derived basal glycolysis, glycolytic capacity, and glycolytic reserve. (F) OCR-derived basal respiration, maximal respiration, and ATP-linked respiration. TEM analyses used n = 5 independent mice per group. Extracellular-flux measurements were obtained using an Agilent Seahorse XF HS Mini Extracellular Flux Analyzer and normalized to total cellular protein. Six independent biological pools were analyzed per group; each pool comprised hippocampi from six non-overlapping mice, corresponding to 36 mice per group. CD11b-enriched cells were seeded at 5.0 × 10^4 cells/well, and three technical replicate wells from each biological pool were averaged before statistical analysis. Each plotted Seahorse data point represents one independent biological pool, not one technical well or one contributing mouse. Data are presented as mean ± SEM. Asterisks indicate comparisons with the Control group: *P < 0.05, ***P < 0.001, and ****P < 0.0001. Number signs indicate comparisons with the untreated PND group: #P < 0.05, ##P < 0.01, and ####P < 0.0001. Because post-enrichment purity was not independently quantified, the preparation is described as a CD11b-enriched cell fraction rather than assigned an unverified numerical microglial purity. In the Mito Stress Test, the CD11b-enriched cell fraction from PND mice showed lower OCR, basal respiration, ATP-linked respiration, and maximal respiration than the corresponding fraction from Control mice (Figure 4D and F). These respiratory measures were higher in the PND + esketamine group than in the untreated PND group, consistent with partial improvement in mitochondrial respiratory function. Taken together, the extracellular flux findings show that esketamine treatment was associated with attenuation of surgery-related bioenergetic abnormalities in the hippocampal CD11b-enriched cell fraction. Because post-enrichment purity was not independently quantified, these data should not be treated as definitive cell-specific evidence and should be interpreted separately from Hk2 and Pkm2 expression measured in whole hippocampal tissue. Six independent biological pools were analyzed per group. Each pool contained hippocampi from six non-overlapping mice, corresponding to 36 mice per group. CD11b-enriched cells were seeded at 5.0 × 10^4 cells per well, and three technical replicate wells from the same pool were averaged before statistical analysis. Systemic 2-DG Is Associated with Selected Behavioral and Day-14 Hippocampal Cytokine Differences To examine whether systemic glycolytic inhibition was associated with selected postoperative outcomes, a contemporaneous five-group cohort was studied according to the timeline shown in Figure 5A. The Control + 2-DG group was randomized and assessed concurrently with the other four groups using the same behavioral and tissue-collection workflow. In the Morris water maze probe trial (Figure 5B), the mean times spent in the target quadrant were 22.40 ± 0.77 s in Control, 11.80 ± 0.55 s in PND, 18.70 ± 0.61 s in PND + esketamine, 19.10 ± 0.93 s in PND + 2-DG, and 22.49 ± 1.64 s in Control + 2-DG. Both intervention groups spent more time in the target quadrant than untreated PND mice (Tukey-adjusted P = 0.0001 for esketamine and P < 0.0001 for 2-DG). Control + 2-DG did not differ from Control (mean difference, 0.09 s; 95% CI, −3.86 to 4.03; adjusted P = 1.0000). The direct PND + esketamine minus PND + 2-DG comparison yielded a mean difference of −0.40 s (95% CI, −4.35 to 3.55; adjusted P = 0.9984). This nonsignificant comparison does not establish equivalence, a shared mechanism, or mediation of esketamine’s effects by glycolytic inhibition. Systemic 2-DG is associated with selected behavioral and day-14 hippocampal cytokine differences in a contemporaneous five-group pharmacological cohort. (A) Experimental timeline. Esketamine (5 mg/kg, i.p.) was administered once after anesthesia induction during surgery on postoperative day 0; 2-DG (500 mg/kg/day, i.p.) was administered to the PND + 2-DG and Control + 2-DG groups once daily on study days 0–2. All five groups were enrolled contemporaneously and underwent the same randomization, blinding, behavioral-testing, and tissue-collection workflow. Morris water maze acquisition was performed on days 8–12, the probe trial on day 13, and hippocampal collection for cytokine assays on day 14. (B) Morris water maze probe-trial outcomes, including representative swimming trajectories and time spent in the target quadrant. The PND + esketamine minus PND + 2-DG mean difference was −0.40 s (95% CI, −4.35 to 3.55; Tukey-adjusted P = 0.9984); Control + 2-DG did not differ from Control (adjusted P = 1.0000). (C) Day-14 hippocampal IL-1β and TNF-α concentrations. The PND + esketamine minus PND + 2-DG mean differences were 4.71 pg/mg protein for IL-1β (95% CI, −8.99 to 18.41; Tukey-adjusted P = 0.8588) and 2.38 pg/mg protein for TNF-α (95% CI, −12.69 to 17.45; Tukey-adjusted P = 0.9908). Control + 2-DG did not differ from Control for either cytokine. (D) Proposed working model integrating results obtained from separate experimental cohorts and time points. In panel D, upward and downward arrows indicate relative increases and decreases, respectively, in the displayed outcomes. Red arrows summarize the directions of change in untreated PND mice relative to Control mice; green arrows summarize the directions of esketamine-associated change relative to untreated PND mice; and blue arrows summarize the selected 2-DG-associated behavioral and day-14 cytokine directions relative to untreated PND mice. Solid connecting arrows summarize observed directions across separate assays, whereas dashed arrows indicate proposed relationships that were not tested causally. Systemic 2-DG is linked only to the measured behavioral and day-14 hippocampal cytokine outcomes and not to unmeasured extracellular-flux effects in hippocampal CD11b-enriched cell fractions. HIF-1α was not measured and is not presented as a study-derived molecular result. Behavioral data are presented as mean ± SEM with n = 10 independent mice per group; day-14 IL-1β and TNF-α data are presented as mean ± SEM with n = 8 predesignated independent mice per group. Asterisks indicate comparisons with the Control group, whereas number signs indicate comparisons with the untreated PND group. *P < 0.05 versus Control; #P < 0.05 versus untreated PND; ns, not significant. Dashed comparison brackets identify direct PND + esketamine versus PND + 2-DG comparisons. The Control + 2-DG group is a contemporaneous non-surgical comparator for the displayed endpoints; it does not establish systemic safety or cell specificity. The nonsignificant direct comparisons do not establish therapeutic equivalence, and the 2-DG observations do not demonstrate that glycolytic inhibition mediates esketamine’s effects. On postoperative day 14, after completion of behavioral testing, hippocampal IL-1β and TNF-α were measured in eight predesignated mice per group (Figure 5C). For IL-1β, the group means were 44.29 ± 2.57 pg/mg protein in Control, 146.99 ± 5.00 pg/mg protein in PND, 71.76 ± 2.41 pg/mg protein in PND + esketamine, 67.05 ± 2.58 pg/mg protein in PND + 2-DG, and 44.13 ± 3.57 pg/mg protein in Control + 2-DG. Compared with PND, IL-1β was lower in both intervention groups (Tukey-adjusted P < 0.0001 for each). The PND + esketamine minus PND + 2-DG mean difference was 4.71 pg/mg protein (95% CI, −8.99 to 18.41; adjusted P = 0.8588). Control + 2-DG did not differ from Control for IL-1β (mean difference, −0.16 pg/mg protein; 95% CI, −13.54 to 13.86; adjusted P = 1.0000). For TNF-α, the group means were 27.12 ± 1.74 pg/mg protein in Control, 82.16 ± 6.03 pg/mg protein in PND, 60.61 ± 3.34 pg/mg protein in PND + esketamine, 58.23 ± 3.66 pg/mg protein in PND + 2-DG, and 30.56 ± 2.17 pg/mg protein in Control + 2-DG. Compared with PND, TNF-α was lower in the PND + esketamine group (mean difference, −21.55 pg/mg protein; 95% CI, −36.61 to −6.48; adjusted P = 0.0020) and the PND + 2-DG group (mean difference, −23.93 pg/mg protein; 95% CI, −38.99 to −8.86; adjusted P = 0.0005). The direct PND + esketamine versus PND + 2-DG comparison yielded a mean difference of 2.38 pg/mg protein (95% CI, −12.69 to 17.45; adjusted P = 0.9908). Control + 2-DG did not differ from Control for TNF-α (mean difference, 3.44 pg/mg protein; 95% CI, −11.63 to 18.50; adjusted P = 0.9644). These day-14 cytokine data are distinct from the early 24-h IL-1β data shown in Figure 3. The principal findings and their evidentiary limits are summarized in the proposed working model (Figure 5D). Because 2-DG was administered systemically and has broad metabolic effects, the observed pattern is compatible with associations between glycolytic inhibition and selected outcomes but does not prove that glycolytic activation caused the phenotype, mediated esketamine’s effects, or acted through hippocampal CD11b-enriched cells. Discussion This study investigated the association between esketamine-mediated neuroprotection and metabolic-inflammatory changes in a tibial fracture model of PND in aged mice. The exploratory analysis of GSE165798 provided an initial hypothesis-generating indication of glycolysis-related and inflammatory alterations, whereas the principal evidence was derived from independent experimental cohorts. At 24 h after surgery, PND mice showed increased hippocampal IL-1β, altered whole-hippocampal glycolysis-related molecular and biochemical markers, and abnormal bioenergetic profiles in hippocampal CD11b-enriched cell fractions; these endpoints were measured in separate early tissue and Seahorse cohorts. Synaptic ultrastructure and behavior were assessed later. In the distinct pharmacological cohort, IL-1β and TNF-α were measured on day 14 after behavioral testing. Esketamine-associated improvements were observed across these separately measured outcomes. Systemic 2-DG was associated with selected behavioral and day-14 cytokine differences, and a contemporaneous Control + 2-DG group was included. Taken together, these findings show a coherent pattern of associations but do not establish a microglia-specific molecular pathway, a causal role for glycolysis, or mechanistic equivalence between 2-DG and esketamine. A central finding was the marked bioenergetic alteration observed in hippocampal CD11b-enriched cell fractions after surgery. Increased ECAR, basal glycolysis, and glycolytic capacity, together with reduced glycolytic reserve, suggest increased glycolytic demand with diminished metabolic flexibility in the enriched fraction. The concurrent reductions in OCR, basal respiration, ATP-linked respiration, and maximal respiration further indicate impaired mitochondrial oxidative function rather than an isolated increase in glycolysis. This combined profile is consistent with the broader concept that inflammatory activation of innate immune cells is frequently accompanied by increased reliance on aerobic glycolysis and constrained mitochondrial respiration.5,21,23–31 Esketamine attenuated both components of this bioenergetic disturbance by limiting excessive glycolytic activation and improving mitochondrial respiratory parameters. These results extend previous observations of the anti-inflammatory and neuroprotective properties of esketamine by showing that its protective phenotype in PND was accompanied by improved bioenergetic function in a hippocampal CD11b-enriched preparation. However, because post-enrichment purity was not independently quantified, the extracellular flux findings establish an association with remodeling in the CD11b-enriched fraction but do not provide definitive microglia-specific evidence or identify the upstream molecular targets responsible for these changes. The whole-hippocampal molecular and biochemical findings provide a complementary, but distinct, level of evidence. Surgery increased the expression of glycolysis-related genes, including Hk2, Ldha, Pdk1, and Pkm2, together with lactate, pyruvate, lactate dehydrogenase activity, and pyruvate kinase activity, whereas esketamine attenuated these changes. HK2 and PKM2 are biologically plausible links between glycolytic activation and inflammatory signaling; for example, PKM2 can influence HIF-1α activity and IL-1β production in activated myeloid cells.32 Nevertheless, the present measurements were obtained from whole hippocampal tissue rather than isolated microglia, and neither Iba1/HK2 nor Iba1/PKM2 colocalization was performed. The supportive public single-cell reference analysis was also derived from a different experimental context. Accordingly, the findings should be interpreted as evidence of hippocampal glycolysis-related signaling that occurred alongside bioenergetic remodeling in hippocampal CD11b-enriched cell fractions, not as proof of a microglia-specific HIF-1α/HK2/PKM2 pathway or direct molecular targets of esketamine. Similarly, qualitative inspection of representative images suggested spatial proximity between Iba1-positive cells and GLUT1-positive microvasculature, but this descriptive observation was not formally quantified and does not demonstrate enhanced glucose uptake, competition for metabolic substrates, a treatment effect on perivascular clustering, or a causal vascular-microglial mechanism. The systemic 2-DG experiment offers an additional pharmacological observation while also defining important limits of mechanistic interpretation. In the contemporaneous five-group cohort, 2-DG was associated with selected differences in spatial memory and day-14 hippocampal IL-1β and TNF-α relative to untreated PND. This pattern is compatible with involvement of glycolysis-related processes, but it does not demonstrate that postoperative glycolytic activation caused the measured phenotype or that inhibiting glycolysis mediated the effects of esketamine. Partial phenotypic overlap also does not show that 2-DG and esketamine act through an identical pathway or have equivalent efficacy. The study was not designed as an equivalence or noninferiority experiment, and nonsignificant treatment-group comparisons do not establish therapeutic equivalence. The Control + 2-DG group improves interpretation of the measured endpoints under a non-surgical condition, but it cannot exclude unmeasured systemic metabolic or long-term effects. Moreover, 2-DG is non-cell-selective and may affect neurons, astrocytes, endothelial cells, peripheral immune cells, and systemic glucose metabolism. Esketamine has broader pharmacological actions and has been associated with NMDA receptor signaling, microglial polarization, autophagy, inflammatory cell-death pathways, and mitochondrial function.33 Differences between the interventions may therefore reflect partially overlapping, parallel, or unrelated mechanisms. From a translational perspective, the findings support further investigation of pathological metabolic-inflammatory coupling as a modifiable component of perioperative brain vulnerability. They do not, however, support nonselective suppression of glycolysis as a clinical strategy. Glycolysis is required for normal function in microglia and multiple other neural, vascular, and peripheral cell populations, so broad inhibition may produce unintended metabolic effects. Esketamine may influence an abnormal postoperative bioenergetic state without functioning as a direct glycolytic inhibitor, but the present animal experiments cannot determine the clinically relevant dose, timing, safety, or cognitive efficacy of such modulation in older surgical patients. Metabolic products such as lactate may also participate in longer-term inflammatory regulation through epigenetic processes, including histone lactylation,34–38 but these mechanisms were not measured here and should remain a focused direction for future study rather than an explanation of the present findings. Several limitations should be acknowledged. First, the exploratory transcriptomic analysis was based on a small public bulk-hippocampal dataset containing three Sham and three PND samples and cannot provide confirmatory or cell-type-specific evidence. Second, Hk2, Pkm2, and the other glycolysis-related molecular and biochemical markers were measured in whole hippocampal tissue; they were not directly quantified in isolated microglia, and no Iba1/HK2 or Iba1/PKM2 colocalization was performed. Third, magnetic enrichment yielded a CD11b-enriched fraction, but post-enrichment purity was not independently quantified by flow cytometry or immunocytochemistry; possible contributions from residual non-microglial cells therefore cannot be excluded. Fourth, extracellular flux in hippocampal CD11b-enriched cell fractions was assessed at a single early postoperative time point using pooled hippocampal samples. Although each independent pool was generated from non-overlapping animals and treated as the statistical unit, pooling limits evaluation of inter-animal variability and does not establish the temporal sequence of metabolic changes. Fifth, the absence of microglia-specific knockdown, conditional knockout, or rescue experiments prevents conclusions that microglial glycolysis, HK2, or PKM2 is required for the effects of esketamine. Sixth, the inflammatory measurements were distributed across distinct cohorts and time points: IL-1β was measured at 24 h in the early tissue cohort, whereas IL-1β and TNF-α were measured on day 14 in the pharmacological cohort. These data cannot define cytokine trajectories or be compared directly across time, and the limited panel did not include other mediators such as IL-6. Seventh, although a contemporaneous Control + 2-DG group was included, the 2-DG cohort assessed only selected behavioral and hippocampal cytokine endpoints; it did not establish cell specificity, brain or systemic metabolic safety, long-term effects, or the absence of off-target actions. Eighth, the Control group accounted for anesthesia and perioperative handling but did not include a separate sham surgical procedure, and only aged male mice were studied, limiting interpretation of procedure-specific and sex-dependent effects. Finally, the sample sizes were modest and the experiments were conducted in a single animal model; consequently, the findings require replication with longitudinal inflammatory measurements, cell-specific approaches, both sexes, broader metabolic and inflammatory profiling, and clinically relevant perioperative studies before they can be extrapolated to human PND prevention or treatment. Conclusion In summary, esketamine attenuated surgery-induced cognitive impairment, the early 24-h hippocampal IL-1β elevation, and synaptic ultrastructural damage in aged mice. In a separate contemporaneous pharmacological cohort assessed on day 14, esketamine and systemic 2-DG were each associated with lower hippocampal IL-1β and TNF-α than untreated PND, while Control + 2-DG provided a non-surgical comparator for the measured endpoints. Across separate cohorts, esketamine-associated protection was accompanied by improved bioenergetic profiles in hippocampal CD11b-enriched cell fractions and reduced expression of glycolysis-related markers, including Hk2 and Pkm2, in whole hippocampal tissue. These findings do not establish that glycolytic activation caused the postoperative phenotype, that HK2 or PKM2 changes occurred specifically in microglia, or that glycolytic inhibition mediated the effects of esketamine. Cell-type-specific and longitudinal studies are needed to establish direct causality. Funding Statement This work was supported by the Science and Technology Project of Sichuan Provincial Health Commission, China (No. 24WSXT050) and the Scientific Research Project of Sichuan Medical Association (No. Q22022, Q2024085). Abbreviations 2-DG, 2-deoxy-D-glucose; ECAR, extracellular acidification rate; GEO, Gene Expression Omnibus; GLUT1, glucose transporter 1; HIF-1α, hypoxia-inducible factor 1-alpha; HK2, hexokinase 2; IL-1β, interleukin-1 beta; TNF-α, tumor necrosis factor alpha; LDH, lactate dehydrogenase; OCR, oxygen consumption rate; PK, pyruvate kinase; PKM2, pyruvate kinase M2; PND, postoperative neurocognitive disorder; qRT-PCR, quantitative real-time polymerase chain reaction; TEM, transmission electron microscopy. Data Sharing Statement The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request. The transcriptomic dataset GSE165798 is publicly available in the NCBI Gene Expression Omnibus repository. RT-qPCR primer sequences are provided in Table S1 in the Supplementary Material. Author Contributions Fei Jiang conceived and supervised the study. Di Liu and Guangchun Li performed the experiments and analyzed the data. Yue Zhang, Hui Huang, and Yong Yang contributed to animal experiments and data collection. Di Liu drafted the manuscript. Fei Jiang revised the manuscript. All authors made a significant contribution to the work reported, whether in the conception, study design, execution, acquisition of data, analysis and interpretation, or in all these areas; took part in drafting, revising, or critically reviewing the article; gave final approval of the version to be published; agreed on the journal to which the article has been submitted; and agree to be accountable for all aspects of the work. Disclosure The authors report no conflicts of interest in this work. References Associated Data Data Availability Statement The datasets generated and analyzed during the current study are available from the corresponding author upon reasonable request. The transcriptomic dataset GSE165798 is publicly available in the NCBI Gene Expression Omnibus repository. RT-qPCR primer sequences are provided in Table S1 in the Supplementary Material.