Desflurane Increases Hippocampal RAF–MAPK–MSK1 Protein Levels and Reduces CLOCK Expression: Sorafenib Attenuates Cognitive Impairment in Aged Mice

Abstract

Postoperative cognitive dysfunction (POCD) is common after anesthesia and surgery in older adults, yet no targeted therapy exists, and the mechanism linking volatile anesthetic exposure to circadian dysregulation in the aged hippocampus is undefined. Here we show that RAF manipulation bidirectionally alters CLOCK protein levels in HT22 hippocampal neurons, and that RAF–MAPK–MSK1 protein abundance is inversely related to CLOCK levels in the aged hippocampus after desflurane exposure. Aged C57BL/6 mice (18 months, both sexes) were exposed to 2% desflurane for 2 h, with or without pretreatment with the multi-kinase RAF inhibitor sorafenib (50 mg/kg), and phenotypes were assessed by Western blot, immunofluorescence, in vivo electroencephalography, whole-cell patch-clamp recordings of CA1 pyramidal neurons, and behavioral testing (righting reflex, contextual fear conditioning, Morris water maze). Parallel dissection in HT22 hippocampal neurons employed RAF knockdown, overexpression, and pharmacological inhibition. Desflurane increased total c-Raf, ERK1/2 (MAPK), and MSK1 protein levels—phosphorylation status was not assessed—while reducing CLOCK in both HT22 cells and aged hippocampus, accompanied by cellular senescence, impaired S-phase entry, increased hippocampal β-galactosidase and α-synuclein immunoreactivity, synaptic excitation–inhibition imbalance, and deficits in contextual fear memory and spatial learning. In HT22 cells, RAF knockdown normalized kinase abundance and restored CLOCK, whereas overexpression exacerbated these changes. In aged mice, sorafenib restored CLOCK, attenuated hippocampal β-galactosidase and α-synuclein immunoreactivity, and improved cognitive performance. These findings support a link between RAF–MAPK–MSK1 protein abundance and hippocampal CLOCK in desflurane-induced cognitive impairment and nominate RAF inhibition as a candidate strategy warranting further preclinical study.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s11064-026-04904-6.

Introduction

Clinical Significance and Emerging Urgency

Worldwide demographic aging has dramatically increased the surgical volume in patients aged 65 years and older [1], in whom postoperative cognitive dysfunction (POCD)—a clinical syndrome characterized by impaired memory, attention, and executive function lasting days to months after anesthesia and surgery—affects up to 36.6% of younger adults and 42.4% of elderly surgical patients [2], with the prevalence ranging from 19% to 64% in non-cardiac surgery and persistent POCD documented in 10–17% of patients at 3 months [3, 4]. POCD is increasingly recognized as a risk factor for premature mortality and accelerated trajectories toward dementia [5]. Despite decades of investigation, POCD remains without an approved targeted therapy, and its clinical management is largely confined to risk identification and supportive care [6]. As perioperative neurocognitive disorders are now formally distinguished from preoperative cognitive baseline and stratified by their temporal relationship to surgery [4], there is mounting pressure to identify the molecular mechanisms that translate brief anesthetic exposure—an intrinsically reversible physiological perturbation—into persistent cognitive decline in the aging brain.

Current Mechanistic Framework of POCD

Mechanistic accounts of POCD have predominantly converged on three interconnected processes: neuroinflammation driven by microglial activation and peripheral cytokine infiltration [7–9], aberrant accumulation of neurodegeneration-associated proteins such as α-synuclein (α-syn) [10, 11], and synaptic excitation–inhibition (E/I) imbalance reflected in altered hippocampal glutamatergic transmission [12]. Volatile anesthetics—including sevoflurane, isoflurane, and desflurane—are widely used in clinical practice and have been shown to disrupt hippocampal synaptic plasticity in aged rodents [13, 14] through mechanisms involving GABA_A receptor potentiation [15], oxidative stress [16], and tau hyperphosphorylation [17, 18]. More recently, attention has shifted toward cell-intrinsic stress-response programs, including cellular senescence and cell-cycle re-entry [19], which have been documented in aged hippocampal neurons and proposed to underlie age-dependent vulnerability to surgical and anesthetic stress [20–22]. In parallel, the circadian transcription factors CLOCK and BMAL1 have emerged as integral regulators of neuronal homeostasis: their expression declines with age, and their disruption accelerates oxidative stress, impairs autophagy, and compromises synaptic pruning [23–25]. These multiple, ostensibly independent pathways suggest that POCD may not result from a single insult but rather from convergent dysregulation of upstream coordinating signals.

Knowledge Gaps

Three substantive gaps in the existing literature constrain mechanistic understanding and therapeutic development for POCD. First, the molecular bridge linking volatile anesthetic exposure to circadian transcription factor dysregulation in the aged hippocampus remains undefined. While BMAL1 deficiency has been mechanistically linked to cognitive aging in numerous models [23, 26], the role of its obligate heterodimeric partner CLOCK in anesthetic-induced cognitive injury has received comparatively little attention. Second, although the RAS–RAF–MEK–ERK (mitogen-activated protein kinase, MAPK) cascade is among the most extensively characterized stress-activated signaling pathways in oncology [27] and is a recognized inducer of cellular senescence in mitotically active cells [28], its potential involvement in volatile anesthetic-induced neuronal stress and POCD pathogenesis has not been systematically investigated. Among the three RAF isoforms, the present study focuses on c-Raf (RAF1; hereafter RAF), which was targeted by our siRNA (against Raf1) and quantified by Western blot as total protein. Recent evidence indicates that mature neurons can re-engage cell-cycle machinery under stress and undergo senescence-like fates [20, 22], suggesting that RAF-driven senescence programs may also operate in post-mitotic neurons of the aged hippocampus. The mitogen- and stress-activated kinase MSK1, a downstream effector of ERK that regulates CREB phosphorylation and histone H3 modification and is required for hippocampal-dependent memory [29], provides a plausible route by which ERK1/2–MSK1 signaling could influence the transcriptional control of clock genes—yet this connection has not been directly tested. Third, the bulk of recent mechanistic work on inhalational anesthetic neurotoxicity has focused on sevoflurane and isoflurane [30], leaving desflurane—a clinically common volatile anesthetic with favorable pharmacokinetics in elderly patients—substantially underexplored at the mechanistic level.

Hypothesis and Study Objectives

To address these gaps, we hypothesized that (i) desflurane increases the protein abundance of the RAF–MAPK–MSK1 cascade in aged hippocampal neurons, (ii) this increase is accompanied by reduced CLOCK expression, and (iii) these molecular changes are associated with convergent cellular and behavioral pathology—including neuronal senescence, impaired proliferation, synaptic E/I imbalance, α-syn accumulation, that is attenuated by pharmacological RAF inhibition. We employed the clinically approved multi-kinase inhibitor sorafenib [31] to test the therapeutic arm of this hypothesis and thereby evaluate whether RAF constitutes a translationally tractable therapeutic node. To test these hypotheses, we used a dual in vitro and in vivo strategy that combines RAF gain- and loss-of-function manipulation in HT22 hippocampal neurons with desflurane exposure in aged C57BL/6 mice, integrating molecular, cellular, electrophysiological, and behavioral readouts.

Materials and methods

Experimental Animals and Ethics Statement

All animal experiments were conducted in strict accordance with the Regulations on the Administration of Laboratory Animals and the Guidelines for the Care and Use of Laboratory Animals, and were approved by the Ethics Committee of the Third Clinical Medical College, Inner Mongolia Medical University (Approval No. 2024-MER-013). Aged C57BL/6 mice (18 months old, both sexes, body weight 32–38 g, both sexes were included to avoid restricting findings to a single sex, consistent with current guidelines for incorporating sex as a biological variable, specific pathogen-free grade) were purchased from the Experimental Animal Center of Inner Mongolia University. Animals were housed under controlled conditions (temperature 20–24 °C; relative humidity 40–65%) with a standard 12 h light/dark cycle (lights on at 08:00, lights off at 20:00), with free access to food and water and minimal noise and light disturbance. All animals were acclimatized for at least one week before any experimental procedure. A total of 54 mice were used in this study. The main cohort comprised 36 mice randomized into three groups (n = 12 per group) using a computer-generated randomization sequence: untreated control, desflurane (Des), and desflurane + RAF inhibitor (Des + RAF inhibitor). After completion of behavioural testing, the 12 animals in each group were allocated to terminal assays as follows: six animals were killed by decapitation, with one hippocampus rapidly dissected and snap-frozen for Western blot and the contralateral hemisphere used to prepare acute slices for whole-cell recording; the remaining six were perfusion-fixed for immunofluorescence. A separate cohort of 18 mice (6 per group, same three treatment groups) underwent stereotaxic electrode implantation for in vivo EEG recording (Sect.  2.12); these animals did not undergo behavioural testing. The group size of n = 12 for the main cohort was based on group sizes used in previous studies of anesthetic-induced cognitive impairment in aged rodents. All behavioural scoring and all image and blot quantification were performed by an investigator blinded to group allocation. No animals were excluded from analysis and no deaths occurred during or after desflurane exposure.

Establishment of the Desflurane Exposure Model

A desflurane inhalation model of anesthetic-induced cognitive impairment was established in aged C57BL/6 mice. No surgical procedure was performed in any animal of the main cohort; the separate EEG cohort underwent electrode implantation as described in Sect. 2.12. Mice were placed individually in a transparent, hermetically sealed anesthesia chamber (maintained at 37 °C on a thermostatic heating pad) and exposed to 2% desflurane (Baxter Healthcare, USA) in 100% oxygen at a flow rate of 2–3 L/min for 2 h. This concentration is below the minimum alveolar concentration (MAC) reported for desflurane in mice [32]. Exposure was delivered without any surgical procedure, an approach used to dissociate anesthetic-specific neuronal stress from surgical trauma [12]. All anesthesia sessions and all terminal procedures were performed during the morning light phase of the 12 h light/dark cycle to ensure consistent circadian conditions across animals [32]. Desflurane concentrations were continuously monitored (Datex-Ohmeda, GE Healthcare, USA), and core body temperature was maintained at 37 ± 0.5 °C. For the Des + RAF inhibitor group, sorafenib (50 mg/kg in 0.5% carboxymethylcellulose; Cat. No. S7397, Selleck Chemicals, USA) was administered by oral gavage 1 h before desflurane exposure. Control mice were placed in the identical chamber and received 100% O₂ (2–3 L/min) for 2 h without desflurane, and were given vehicle only.

HT22 Cell Culture, Desflurane Exposure, and Experimental Grouping

HT22 cells (hippocampal neuronal line) were maintained in DMEM with 10% FBS (Gibco, USA) and 1% penicillin–streptomycin at 37 °C, 5% CO₂, and subcultured at ~ 85% confluence using 0.25% trypsin (Gibco, Cat. No. 25200-072). For desflurane exposure, cells were placed in a temperature-controlled airtight chamber (37 °C, 5% CO₂) and exposed to 2.0% desflurane for 2 h. This concentration produced a submaximal reduction in viability (Fig. 1B) while retaining sufficient viable cells for downstream assays. Seven experimental groups were used (n = 3/group): Control, Des, Des + si-NC, Des + si-RAF, Des + oe-NC, Des + oe-RAF, and Des + RAF inhibitor (10 µM sorafenib, added 1 h prior to desflurane).

Fig. 1.

RAF regulates MAPK–MSK1 cascade protein levels and CLOCK abundance in desflurane-exposed HT22 hippocampal neurons, and bidirectionally regulates neuronal viability. A Cell viability across groups by CCK-8 assay. Desflurane reduced viability to ~ 67% (P < 0.001 vs. Control); RAF knockdown and pharmacological inhibition partially restored viability, whereas RAF overexpression aggravated the reduction. B Concentration–response relationship of desflurane (0–3.0%) on HT22 cell viability (CCK-8 assay). Cell survival remained largely stable at ≤ 1.0%, declined sharply to ~ 60% at 1.5–2.0%, and fell below 50% at ≥ 2.5% (P < 0.0001 for 2.0–3.0% vs. 0%). Based on this profile, 2.0% (dashed line) was selected as the working concentration. C RT-qPCR-based screening of three RAF-targeting siRNA sequences (siRNA1, siRNA2, siRNA3). siRNA1 achieved the highest knockdown efficiency, reducing RAF mRNA to approximately 45% of the si-NC control (P < 0.0001 vs. si-NC; P < 0.01 vs. siRNA2 and siRNA3), and was therefore selected for all subsequent knockdown experiments. D RT-qPCR confirmation of RAF overexpression efficiency. Oe-RAF increased RAF mRNA ~ 1.9-fold compared with Oe-NC (P < 0.001). E Representative Western blot images of ERK1/2 (43 kDa), RAF (74 kDa), MSK1 (90 kDa), and CLOCK (100 kDa) across all seven experimental groups; β-actin (43 kDa) served as the loading control. F–I Densitometric quantification of RAF (F), ERK1/2 (G), MSK1 (H), and CLOCK (I) normalized to β-actin and expressed relative to the Control group (set to 1.0). Desflurane significantly upregulated RAF (P < 0.0001), ERK1/2 (P < 0.0001), and MSK1 (P < 0.001) while significantly suppressing CLOCK (P < 0.01) relative to controls. RAF knockdown and pharmacological inhibition each reversed these changes; conversely, RAF overexpression further increased RAF, ERK1/2, and MSK1 protein levels (all P < 0.01–0.001 vs. oe-NC) and maximally suppressed CLOCK (~ 0.23; P < 0.001 vs. oe-NC). Data are mean ± SD; n = 3 independent experiments. *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; one-way ANOVA with Tukey’s multiple comparisons test

siRNA Knockdown and Plasmid Overexpression of RAF

One day before transfection, HT22 cells were seeded into 6-well plates at a density of ≥ 2 × 10⁵ cells/well. Three RAF-targeting siRNA duplexes (siRNA1–3) and a non-targeting si-NC were designed against the mouse Raf1 coding sequence (NM_029780) and synthesized by Sangon Biotech (Shanghai, China); full sequences are available upon reasonable request. siRNA1 achieved the highest knockdown efficiency (~ 55% reduction; P < 0.0001 vs. si-NC) and was used for all loss-of-function experiments. siRNA (100 nM) was transfected with Lipofectamine™ 3000 (Thermo Fisher, Cat. No. L3000001) in serum-free DMEM following the manufacturer’s protocol. For overexpression, 2 µg of oe-RAF or oe-NC plasmid was transfected using the same protocol. Cells were washed and medium replaced with serum-free DMEM before transfection. After 6 h, transfection medium was replaced with complete growth medium, and cells were harvested 48 h later for RNA and protein analyses. Knockdown and overexpression efficiencies were confirmed by RT-qPCR (Sect.  2.5) and Western blot (Sect.  2.6).

RNA Extraction and RT-qPCR

Total RNA was extracted from HT22 cells or hippocampal tissue using TRIzol (Aidlab, Cat. No. RN0102) per the manufacturer’s protocol, with chloroform phase separation and isopropanol precipitation. RNA concentration and purity were assessed spectrophotometrically (NanoDrop Unano-1000, Youmi Instruments); only samples with A260/A280 ratios between 1.8 and 2.1 were used. First-strand cDNA was synthesized from 1 to 2 µg total RNA using the ExonScript RT SuperMix with dsDNase kit (EXONGEN, Cat. No. A502) at 25 °C for 10 min, 55 °C for 15 min, and 85 °C for 5 min.

RT-qPCR reactions (20 µL) contained 1 µL cDNA, 0.4 µL each of forward and reverse primers (10 µM), 10 µL SYBR Green Master Mix (Applied Biosystems, Cat. No. 4472920), and nuclease-free water to volume, run on a Longgene Q2000B real-time PCR system. Cycling conditions: 95 °C for 5 min; 40 cycles of 95 °C for 10 s, 58 °C for 20 s, 72 °C for 20 s; followed by a melt-curve analysis (95 °C for 15 s, 60 °C for 60 s, 95 °C for 15 s) to confirm amplicon specificity. Relative expression was calculated by the 2^−ΔΔCt method with β-actin as the reference gene. Primer sequences: β-actin forward 5′-CTCCTGAGCGCAAGTACTCT-3′, reverse 5′-TACTCCTGCTTGCTGATCCAC-3′; RAF (Raf1) forward 5′-TACTCGGATCGGGTCAGGTT-3′, reverse 5′-ACCCCATGAACAGCAGGATG-3′.

Western Blot Analysis

Total protein was extracted in RIPA buffer (Beyotime, Cat. No. P0013C) with protease inhibitor cocktail (Beyotime, Cat. No. P1005), centrifuged at 12,000 × g for 10 min at 4 °C, and quantified by BCA assay (Aoxing, Cat. No. AX913053). Equal protein (30 µg/lane) was resolved on 10% SDS-PAGE and transferred onto 0.22 μm nitrocellulose membranes (GVS, Cat. No. ISEQ00010). Membranes were blocked in 5% non-fat skimmed milk/1× TBST for 1 h at room temperature, then incubated overnight at 4 °C with the following primary antibodies (all from Affinity Biosciences, China unless stated): anti-c-Raf/RAF1 (1:1000, Cat. No. AF6065; detects total c-Raf), anti-ERK1/2 (p44/42 MAPK) (1:1000, Cat. No. AF6456), anti-MSK1 (1:1000, Cat. No. AF6031), anti-CLOCK (1:1000, Cat. No. P9436), and anti-β-actin (1:3000, Cat. No. AF7018). After three 10-min washes with 1× TBST, membranes were incubated with HRP-conjugated secondary antibodies (goat anti-rabbit IgG or goat anti-mouse IgG, 1:3000; Saierbio, China) for 1 h at room temperature. Bands were visualized by enhanced chemiluminescence (ECL; CLINX, China) using a ChemiScope 6100 imaging system (CLINX, China). Band densities were quantified with ImageJ (NIH, USA) and normalized to β-actin. Because ERK1/2 and β-actin have closely similar molecular weights and cannot be resolved on a single membrane, they were detected on separate membranes loaded with the same samples in the same order.

Immunofluorescence Staining

Mice were perfused with cold PBS then 4% PFA; brains were post-fixed (overnight, 4 °C), paraffin-embedded, and sectioned (4 μm). Sections were deparaffinized, rehydrated, and subjected to high-pressure antigen retrieval (2 min). Following PBS washes (3 × 2 min), sections were blocked with 5% normal goat serum in PBS at 37 °C for 60 min, then incubated overnight at 4 °C with anti-α-synuclein primary antibody (α-syn; 1:200; rabbit polyclonal; Affinity Biosciences, Cat. No. AF6285). After PBS washes (5 × 3 min), sections were incubated for 1 h at 37 °C in the dark with Alexa Fluor 488-conjugated goat anti-rabbit IgG (H + L) secondary antibody (1:500; Thermo Fisher Scientific, Cat. No. A11008). β-galactosidase signal in hippocampal sections was detected using the Senescence-Tracker™ NIR fluorescent probe as described in Sect.  2.10. Nuclei were counterstained with DAPI (Beyotime, Cat. No. C1005), sections were washed with PBS, and coverslips were mounted with anti-fade medium (Aoxing, Cat. No. 30093360). Images were acquired on a fluorescence microscope (Nikon ECLIPSE-CI; 20× objective), and fluorescence intensity was quantified using ImageJ (version 1.53, NIH, USA).

Cell Viability Assay (CCK-8)

Cell viability was assessed by CCK-8 assay (Fubaike Biotechnology, Beijing, Cat. No. DCM7122). HT22 cells were seeded in 96-well plates (3 × 10⁴ cells/well) 24 h before treatment. After treatments, 10 µL CCK-8 solution (Fubaike Biotechnology, Cat. No. DCM7122) was added per well, incubated at 37 °C for 1 h, and absorbance at 450 nm was measured (Bio-Rad Model 550); wells without cells served as blanks. All measurements were performed in triplicate; relative viability (%) was calculated as (ODsample − ODblank) / (ODcontrol − ODblank) × 100%. For concentration-response experiments, cells were exposed to desflurane at 0%, 0.5%, 1.0%, 1.5%, 2.0%, 2.5%, and 3.0% for 2 h, and CCK-8 viability was measured 24 h post-exposure.

EdU Incorporation Assay for Cell Proliferation

HT22 cells on coverslips received EdU (10 µM final; Beyotime, Cat. No. C0071S) for 2 h at 37 °C, then were fixed with 4% PFA (RT, 15 min), washed, and permeabilized (10–15 min). After washing (1–2 × 5 min), 0.5 mL Click reaction solution was added per well and incubated in the dark at room temperature for 30 min. After three washes (5 min each), nuclei were counterstained with DAPI, coverslips mounted, and images captured under a fluorescence microscope (Nikon ECLIPSE-CI; green channel for EdU). Relative EdU fluorescence intensity was quantified using ImageJ.

Senescence-Associated β-Galactosidase Staining

Senescence-associated β-galactosidase signal was detected using the Senescence-Tracker™ NIR fluorescent probe (Beyotime, Cat. No. C0603), which specifically detects β-galactosidase activity at pH 6.0. For HT22 cells, cells were incubated with the probe (10 µM working solution in complete medium) at 37 °C for 24 h, washed three times with PBS, and imaged on a fluorescence microscope (Nikon ECLIPSE-CI; red/NIR channel). For in vivo experiments, paraffin-embedded hippocampal sections (prepared as described in Sect.  2.7) were deparaffinized, rehydrated, and incubated with the same probe (10 µM in PBS, pH 6.0) at 37 °C for 4 h in the dark, washed three times with PBS, mounted with anti-fade medium containing DAPI, and imaged under the red/NIR channel. Fluorescence intensity was quantified using ImageJ.

Cell Cycle Analysis by Flow Cytometry

Cell cycle distribution was determined by propidium iodide (PI) DNA staining. After indicated treatments, HT22 cells were harvested by trypsinization, washed twice with ice-cold PBS, and fixed in 70% ethanol at − 20 °C overnight. Fixed cells were washed twice with PBS, incubated with RNase A (100 µg/mL in PBS) at 37 °C for 30 min to degrade RNA, then stained with PI (50 µg/mL in PBS; Beyotime) in the dark at room temperature for 30 min. Stained cells were analyzed on a flow cytometer (BD FACSCanto II), collecting ≥ 10,000 events per sample. Cell cycle phase proportions (G0/G1, S, and G2/M) were determined using ModFit LT 5.0 (Verity Software House) or FlowJo v10 software. All experiments were performed in triplicate (n = 3).

In Vivo Electroencephalography (EEG) Recording

In vivo EEG recordings were performed in aged C57BL/6 mice. Mice were anesthetized with isoflurane (4% induction, 1.5–2% maintenance in O₂) with buprenorphine (0.1 mg/kg, s.c.) analgesia and fixed in a stereotaxic apparatus (RWD Life Science, China) on a 37 °C heating pad. Bilateral hippocampal CA1 recording electrodes (− 2.0 mm A/P, ± 1.5 mm M/L, − 1.5 mm D/V from bregma) and a cerebellar screw serving as reference/ground were implanted and secured with dental cement; the identical procedure and anesthetic regimen were applied to all groups. After a 3-day recovery, mice in the Des and Des + RAF inhibitor groups received the same desflurane exposure described in Sect.  2.2 (2% desflurane in 100% O₂, 2–3 L/min, 2 h), while Control mice received 100% O₂ alone for 2 h; the electrode assembly remained in place but disconnected during exposure. Twenty-four hours after the end of exposure, spontaneous hippocampal network activity was recorded for 30 min in the awake resting state using an AD Instruments PowerLab acquisition system (LabChart software); signals were digitized at ≥ 1 kHz and band-pass filtered (1–100 Hz), with a 50-Hz notch filter. Oscillatory bursts were detected from the Hilbert amplitude envelope of the filtered signal as events exceeding 2 SD above the mean lasting ≥ 3 oscillatory cycles; burst frequency was expressed as bursts per minute and computed in custom MATLAB scripts. For the Des + RAF inhibitor group, sorafenib (50 mg/kg) was administered by oral gavage 1 h before the onset of desflurane exposure, as described in Sect.  2.2. Because electrode implantation required isoflurane anesthesia in all three groups, animals in this cohort, including Control mice, were not anesthetic-naive at the time of recording.

Acute Brain Slice Preparation and Whole-Cell Patch-Clamp Recordings

Acute coronal hippocampal brain slices (300 μm) were prepared following rapid decapitation from the same six mice per group used for Western blot; slices were cut from the hemisphere contralateral to the one snap-frozen for protein extraction. Brains were dissected in ice-cold NMDG-based cutting solution and sectioned on a vibratome (Leica VT1200S). Slices were transferred to oxygenated ACSF at room temperature and allowed to recover for 30–90 min.

Whole-cell patch-clamp recordings were obtained from visually identified hippocampal CA1 pyramidal neurons using an upright microscope with infrared-DIC optics and a Multiclamp 700B amplifier (Molecular Devices, USA). Patch pipettes (3–6 MΩ, borosilicate glass) were filled with the following intracellular solutions. For mEPSC recordings (in mM): 130 CsMeSO₃, 10 HEPES, 11 EGTA, 1 MgCl₂, 1 CaCl₂, 2 Mg-ATP, 0.3 Na₃-GTP (pH 7.3; ~290 mOsm). For mIPSC recordings (in mM): 140 CsCl, 10 HEPES, 0.2 EGTA, 4 Mg-ATP, 0.3 Na-GTP (pH 7.3 with CsOH; 290 mOsm). Slices were perfused continuously with oxygenated ACSF at 2 mL/min. All mEPSC and mIPSC recordings were performed with 1 µM tetrodotoxin (TTX; Tocris) to isolate action-potential-independent quantal events. Specifically, mEPSCs were pharmacologically isolated by bath application of 100 µM picrotoxin (PTX; GABA_A receptor antagonist) plus 1 µM TTX; mIPSCs were isolated by 10 µM CNQX (AMPA/kainate receptor antagonist), 50 µM D-AP5 (NMDA receptor antagonist), and 1 µM TTX. All recordings were performed at a holding potential of − 70 mV in voltage-clamp gap-free mode for 5 min. Cells showing > 20% change in series resistance or series resistance > 30 MΩ were excluded. Signals were acquired at 10 kHz, low-pass filtered at 2 kHz, and analyzed using Clampfit 10.7 (Molecular Devices) or Mini Analysis Software (Synaptosoft).

Behavioral Testing

Loss-of-righting reflex (LORR) and recovery-of-righting reflex (RORR). To evaluate anesthetic sensitivity and recovery, mice were placed individually in a transparent airtight chamber and exposed to 2% desflurane in 100% O₂. LORR was the time from desflurane onset to loss of righting (confirmed by 3 consecutive failed attempts at 10-s intervals); RORR was the time from desflurane cessation to recovery of righting (3 successful attempts). All observations were made by a trained observer blinded to group assignment.

Contextual fear conditioning (CFC). Contextual fear memory was assessed using a two-phase paradigm. On Day 1, mice freely explored the conditioning chamber for 3 min, then received three CS–US pairings (tone, 80 dB, 30 s; footshock, 0.5 mA, 2 s; 60-s ITI). Contextual memory was tested at 24 h (Day 2) and 48 h (Day 3) by returning mice to the original context for 5 min without any stimulus; percentage freezing (% Freezing, defined as the complete absence of movement except respiration) was recorded by automated video-tracking software.

Morris water maze (MWM). Spatial learning and memory were assessed in a circular pool (diameter 120 cm, depth 30 cm, water temperature 25 ± 1 °C, opacified with non-toxic white paint). A circular escape platform (diameter 10 cm) was submerged 1 cm below the water surface in the target quadrant. During acquisition (Days 1–5), mice received 4 trials/day (120 s max, ITI ≥ 15 min); failed mice were guided to the platform for 10 s. On Day 6, a 120-s probe trial was conducted with the platform removed. All behavioral data were acquired and analyzed using PanLab Smart 3 software; swim speed and path length were also recorded.

Statistical Analysis

Statistical analyses were performed with SPSS 22.0 and GraphPad Prism 9.0; data are mean ± SD. Normality (Shapiro–Wilk) and variance homogeneity (Levene’s) were assessed prior to testing. For multi-group comparisons, one-way ANOVA followed by Tukey’s multiple comparisons post hoc test was applied. Repeated-measures behavioral data (e.g., MWM acquisition) were analyzed by two-way RM-ANOVA (Greenhouse–Geisser correction when sphericity was violated) with Bonferroni-corrected post hoc comparisons. For two-group comparisons, an unpaired two-tailed Student’s t-test was used. Statistical significance was defined as two-sided P < 0.05. Associations between protein levels and cellular readouts were assessed by Spearman’s rank correlation, with Pearson’s r reported in parallel; correlations used paired per-sample values (n = 21; 3 independent biological replicates × 7 experimental conditions, assayed in matched batches).

Results

Desflurane Increases RAF–MAPK–MSK1 Protein Levels and Reduces CLOCK Expression in HT22 Cells

HT22 cells exposed to 0–3.0% desflurane showed dose-dependent viability decline; 2.0% was selected as the working concentration for subsequent assays (Fig. 1B).

At this working concentration, desflurane significantly reduced HT22 cell viability relative to the control group (P < 0.001, Fig. 1A). Western blot analysis further showed that desflurane exposure was accompanied by increased RAF, ERK1/2, and MSK1 protein levels, together with reduced CLOCK expression (all P < 0.05, Fig. 1E–I). To examine whether RAF is causally involved, RAF was manipulated by siRNA knockdown or plasmid-mediated overexpression. siRNA1 achieved ~ 55% RAF mRNA knockdown (Fig. 1C), while overexpression yielded ~ 1.9-fold induction (Fig. 1D; both P < 0.001). RAF protein changes paralleled mRNA (Fig. 1E, F). RAF knockdown attenuated desflurane-induced ERK1/2 and MSK1 elevation and partially restored CLOCK, whereas overexpression amplified these changes (Fig. 1G–I), supporting a causal role for RAF in the accompanying reduction of CLOCK.

Modulation of RAF Affects HT22 Cell Proliferation, Senescence, and Cell-Cycle Progression Under Desflurane Exposure

We next examined whether RAF modulation influenced injury-related cellular phenotypes. Desflurane markedly reduced EdU-positive HT22 cells; RAF knockdown/inhibition partially restored EdU incorporation, while RAF overexpression further reduced it (P < 0.001, Fig. 2A, C).

Fig. 2.

RAF silencing and pharmacological inhibition rescue desflurane-induced proliferative arrest, while RAF overexpression exacerbates neuronal senescence in HT22 cells. A, C Representative IF images (A) and quantification (C) of EdU incorporation (green) with DAPI (blue) in HT22 cells. Desflurane suppressed EdU signal (~ 0.18 vs. ~0.44; P < 0.0001 vs. Control); RAF knockdown and inhibition partially restored (~ 0.27 and ~ 0.23; P < 0.0001), while RAF overexpression further reduced incorporation (~ 0.13; P < 0.001 vs. Des + oe-NC). B, D Representative immunofluorescence images (B) and quantification of senescence-associated β-galactosidase (SA-β-gal; red channel) fluorescence intensity (D). Desflurane significantly elevated SA-β-gal fluorescence compared with untreated controls (~ 0.14 vs. ~0.05; P < 0.0001), indicating pronounced cellular senescence. RAF knockdown markedly attenuated this pro-senescent effect (~ 0.07 vs. ~0.14 in Des + si-NC; P < 0.0001 vs. Des + si-NC). RAF inhibitor treatment also reduced SA-β-gal fluorescence (~ 0.10; P < 0.05 vs. Des), though levels remained above those of the untreated control. Conversely, RAF overexpression substantially exacerbated senescence (~ 0.23; P < 0.0001 vs. Des + oe-NC), confirming the pro-senescent role of RAF signaling. Scale bar: 50 μm. Data are mean ± SD; n = 3 independent experiments. *P < 0.05, ***P < 0.001, ****P < 0.0001; one-way ANOVA with Tukey’s multiple comparisons test

Senescence-associated β-gal staining showed parallel changes: desflurane elevated β-gal activity; RAF knockdown/inhibition attenuated, while overexpression further enhanced it (P < 0.001, Fig. 2B, D). To further characterize the basis of the proliferation defect, cell-cycle distribution was analyzed by flow cytometry. Desflurane reduced the S-phase cell fraction, indicating impaired cell-cycle progression (Fig. 3A–H). RAF knockdown or pharmacological inhibition increased S-phase entry (P < 0.01), whereas RAF overexpression further reduced it (P < 0.05, Fig. 3H), supporting a RAF-dependent regulation of cell-cycle progression.

Fig. 3.

RAF overexpression reduces S-phase entry in desflurane-exposed HT22 neurons; RAF inhibition partially restores S-phase entry. A–G Representative propidium iodide (PI) staining flow cytometry histograms displaying DNA content distribution across three cell cycle phases (G1/2 N, S, and G2-M/4 N) for all seven experimental groups: Control (A), Des (B), Des + si-NC (C), Des + si-RAF (D), Des + oe-NC (E), Des + oe-RAF (F), and Des + RAF inhibitor (G). H Quantification of S-phase cell proportion. The Control group showed an S-phase fraction of ~ 38.3%. Desflurane significantly reduced the S-phase proportion (~ 30.0%; P < 0.001 vs. Control), indicating G1-phase arrest and impaired DNA synthesis. Notably, si-NC (~ 26.1%) and oe-NC (~ 18.6%) further reduced S-phase, likely reflecting transfection stress. RAF knockdown restored S-phase entry (~ 37.0%; P < 0.05 vs. Des + si-NC), pharmacological RAF inhibition partially restored it (~ 33.0%), and RAF overexpression maximally reduced it (~ 13.2%; P < 0.05 vs. Des + oe-NC). Data are mean ± SD; n = 3 independent experiments. *P < 0.05, ***P < 0.001, ****P < 0.0001; one-way ANOVA with Tukey’s multiple comparisons test

Desflurane-Induced Increases in RAF–MAPK–MSK1 Protein Levels and CLOCK Reduction are Reproduced In Vivo and Accompanied by Hippocampal Senescence-Related Changes

Hippocampal tissues from aged mice were analyzed by Western blot to test in vivo recapitulation. Compared with controls, desflurane exposure was associated with increased RAF, ERK1/2, and MSK1 protein levels and reduced CLOCK expression in the hippocampus (all P < 0.05, Fig. 4A–E). Treatment with a RAF inhibitor partially reversed these changes, lowering RAF, ERK1/2, and MSK1 levels while increasing CLOCK expression relative to the desflurane group. The directionality of these in vivo changes was consistent with the in vitro findings in HT22 cells (Fig. 1F–I).

Fig. 4.

Desflurane increases RAF–MAPK–MSK1 protein levels and suppresses CLOCK expression in hippocampal tissue of aged mice; RAF inhibition reverses these alterations. A Representative Western blot images of ERK1/2 (43 kDa), RAF (74 kDa), MSK1 (90 kDa), and CLOCK (100 kDa) in hippocampal tissue lysates from Control, Des, and Des + RAF inhibitor groups; β-actin (43 kDa) served as the loading control. Note: because ERK1/2 and β-actin share similar molecular weights, these proteins were detected on separate membranes to ensure specificity. B–E Densitometric quantification normalized to β-actin and expressed relative to the Control group (set to 1.0). Desflurane upregulated RAF (~ 1.35; P < 0.001), ERK1/2 (~ 1.38; P < 0.001), MSK1 (~ 1.25; P < 0.0001), and downregulated CLOCK (~ 0.58; P < 0.001 vs. Control). RAF inhibition reversed all kinase elevations (P < 0.0001 vs. Des) and partially restored CLOCK (~ 0.82; P < 0.05 vs. Des). Data are mean ± SD; n = 6 mice per group. *P < 0.05, ***P < 0.001, ****P < 0.0001; one-way ANOVA with Tukey’s multiple comparisons test

Immunofluorescence showed increased hippocampal β-galactosidase and α-synuclein signal intensity after desflurane exposure; both were attenuated by RAF inhibition (P < 0.01, Fig. 5A–D).

Fig. 5.

RAF inhibition attenuates desflurane-induced hippocampal senescence-associated and α-synuclein immunoreactivity changes in aged mice. A, C Representative immunofluorescence images (A) and quantification (C) of senescence-associated β-galactosidase (SA-β-gal; red channel) co-stained with DAPI (nuclear counterstain; blue) in hippocampal sections. Desflurane exposure markedly increased SA-β-gal fluorescence intensity compared with controls (~ 0.14 vs. ~0.063; P < 0.0001), consistent with a senescence-associated increase in hippocampal β-galactosidase signal. RAF inhibitor treatment significantly attenuated this effect (~ 0.105; P < 0.001 vs. Des; P < 0.001 vs. Control), though fluorescence remained above the control level. B, D Representative IF images (B) and quantification (D) of α-synuclein (green). Desflurane elevated α-syn fluorescence (~ 0.19 vs. ~0.07; P < 0.0001 vs. Control), and RAF inhibition partially attenuated this increase (~ 0.14; P < 0.01 vs. Des). Scale bar: 50 μm. Data are mean ± SD; n = 6 mice per group. **P < 0.01, ***P < 0.001, ****P < 0.0001; one-way ANOVA with Tukey’s multiple comparisons test

RAF Inhibition Partially Restores Synaptic Function Altered by Desflurane Exposure

In vivo hippocampal EEG recordings obtained 24 h after exposure showed that spontaneous oscillatory discharge frequency was significantly reduced in desflurane-exposed mice (P < 0.0001, Fig. 6A–B).

Fig. 6.

RAF inhibition partially restores hippocampal neuronal firing activity and corrects desflurane-induced excitatory/inhibitory synaptic imbalance. A, B Representative in vivo hippocampal EEG traces (A) and quantification of spontaneous network oscillatory burst frequency (B) recorded 24 h after the end of exposure in the awake resting state from a separate implanted cohort of Control, Des, and Des + RAF inhibitor mice. Desflurane significantly reduced burst frequency compared with controls (~ 32 vs. ~52 bursts/min; P < 0.0001). Co-treatment with a RAF inhibitor partially restored oscillatory activity (~ 43 bursts/min; P < 0.01 vs. Des; P < 0.01 vs. Control), indicating incomplete but significant recovery. C–E Representative mEPSC traces (C) and quantification of amplitude (D) and frequency (E). Desflurane reduced mEPSC amplitude (~ 11 vs. ~21 pA; P < 0.001) and frequency (~ 1.2 vs. ~5.0 events/s; P < 0.001 vs. Control). RAF inhibition partially restored both: amplitude ~ 17 pA (P < 0.01 vs. Des), frequency ~ 2.8 events/s (P < 0.05 vs. Des). F–H Representative miniature inhibitory postsynaptic current (mIPSC) traces (F) and quantification of mIPSC amplitude (G) and frequency (H). In contrast to mEPSCs, desflurane significantly elevated mIPSC amplitude (~ 18 vs. ~10 pA; P < 0.001) and frequency (~ 4.9 vs. ~2.8 events/s; P < 0.001 vs. Control), indicating enhanced GABAergic inhibition. RAF inhibitor treatment substantially normalized these parameters: amplitude was reduced to ~ 15 pA (P < 0.05 vs. Des; remaining elevated vs. Control, P < 0.01), and frequency was restored to ~ 3.0 events/s, approaching Control levels (P < 0.001 vs. Des; not significantly different from Control). Data are mean ± SD. For A–B, n = 6 mice per group (separate implanted cohort, recorded 24 h after exposure); for C–H, n = 8–10 neurons per group recorded from six mice per group of the main cohort at the terminal time point (one to two neurons per animal, from ≥ 3 independent slices). Scale bars: 20 pA/2 s (mEPSC); 10 pA/2 s (mIPSC). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001; one-way ANOVA with Tukey’s multiple comparisons test

Whole-cell patch-clamp recordings further showed that desflurane altered both excitatory and inhibitory synaptic transmission. Miniature excitatory postsynaptic currents (mEPSCs) showed reduced amplitude and frequency in the desflurane group relative to controls (P < 0.001, Fig. 6C–E), indicating weakened excitatory synaptic transmission. In contrast, miniature inhibitory postsynaptic currents (mIPSCs) exhibited increased amplitude and frequency after desflurane exposure (P < 0.001, Fig. 6F–H), consistent with enhanced inhibitory synaptic input. RAF inhibition partially reversed both changes, indicating that desflurane disrupts hippocampal excitatory–inhibitory balance and that RAF inhibition attenuates this imbalance.

RAF Inhibition Attenuates Desflurane-Induced Behavioral Abnormalities in Aged Mice

To evaluate the behavioral relevance of the above findings, aged mice were subjected to a series of behavioral tests following desflurane exposure. In the LORR/RORR paradigm, desflurane-exposed mice exhibited shorter LORR times and longer RORR times than mice receiving the RAF inhibitor (P < 0.001 and P < 0.0001, respectively, Fig. 7A–B). Relative to the Des group, RAF inhibitor pretreatment prolonged LORR and shortened RORR, indicating that RAF inhibition is associated with a delayed onset of, and a faster recovery from, desflurane-induced loss of righting.

Fig. 7.

RAF inhibition ameliorates desflurane-induced multi-domain cognitive impairment and modulates anesthetic kinetics in aged mice. A–B. Individual-animal heatmap (A) and group-level quantification (B) of loss-of-righting reflex (LORR) induction time and recovery-of-righting reflex (RORR) time in Des and Des + RAF inhibitor groups. RAF inhibitor treatment significantly prolonged LORR time (P < 0.001) and shortened RORR time (P < 0.0001) compared with the Des group, indicating that RAF inhibitor treatment was associated with a delayed onset of, and a faster recovery from, loss of righting. C–D. Percentage freezing (% Freezing) in contextual fear memory at 24-h (C) and 48-h (D) recall. Desflurane reduced freezing at both time points (24 h: ~34% vs. ~62%; 48 h: ~38% vs. ~69%; P < 0.0001 vs. Control). RAF inhibition substantially restored freezing (24 h: ~55%; 48 h: ~60%; P < 0.0001 vs. Des). E. Morris water maze (MWM) hidden-platform escape latency during 5-day acquisition training. Groups did not differ significantly on Days 1–2, but Des-exposed mice showed significantly prolonged escape latency vs. both Control and Des + RAF inhibitor groups on Days 3, 4, and 5 (***P < 0.001 at each time point), indicating impaired spatial learning that was substantially attenuated by RAF inhibition. F. Probe trial (platform removed): percentage of time spent in the target quadrant. Desflurane significantly reduced target quadrant occupancy compared with controls (~ 25% vs. ~39%; P < 0.0001), indicating impaired spatial memory retention. RAF inhibitor treatment partially restored target quadrant preference (~ 36%; P < 0.0001 vs. Des). Data are mean ± SD; n = 12 mice per group. *P < 0.05, ***P < 0.001, ****P < 0.0001; unpaired two-tailed Student’s t-test (B); one-way ANOVA with Tukey’s multiple comparisons test (C, D, F); two-way repeated-measures ANOVA with Bonferroni post hoc correction (E)

In CFC, desflurane reduced freezing at 24 h and 48 h, an effect substantially reversed by RAF inhibition (all P < 0.0001, Fig. 7C–D). Spatial learning and memory were further assessed using the Morris water maze. During hidden-platform training, desflurane-exposed mice showed longer escape latencies than controls, particularly from the later training days onward (P < 0.001, Fig. 7E). In the probe trial, desflurane exposure reduced target quadrant occupancy, whereas RAF inhibitor treatment partially restored this spatial preference (P < 0.0001 vs. desflurane, Fig. 7F). Swim speed was comparable across groups (Control 18.2 ± 2.1, Des 20.1 ± 3.8, Des + RAF inhibitor 19.3 ± 3.1 cm/s; P > 0.05), indicating that escape-latency differences reflect spatial learning rather than locomotor impairment.

CLOCK Levels are Inversely Correlated with RAF–MAPK–MSK1 Protein Abundance and with Senescence Across Experimental Conditions

To examine whether the kinase–CLOCK relationship held across the full range of experimental conditions, correlation analyses were performed on paired per-sample values from all seven HT22 groups. CLOCK protein levels were inversely correlated with RAF (Spearman r = − 0.62, P = 0.002), ERK1/2 (r = − 0.71, P = 0.0003) and MSK1 (r = − 0.76, P < 0.0001; Fig. 8A–C). CLOCK was also inversely correlated with SA-β-gal signal (r = − 0.72, P = 0.0002) and positively correlated with EdU incorporation (r = + 0.71, P = 0.0003; Fig. 8D–E), while the three kinases were strongly intercorrelated (Fig. 8F). These associations are consistent with the proposed kinase–CLOCK–senescence relationship, although correlation across conditions does not by itself establish causality.

Fig. 8.

CLOCK abundance correlates inversely with RAF–MAPK–MSK1 protein levels and with senescence across experimental conditions in HT22 cells. A–C Correlation between CLOCK and RAF (A), ERK1/2 (B) and MSK1 (C) protein levels, quantified as described in Sect.  2.6 and normalized to β-actin; for ERK1/2, normalization used the β-actin signal from the parallel membrane loaded with the same samples in the same order. D, E Correlation between CLOCK and SA-β-gal fluorescence intensity (D) and EdU incorporation (E). F Correlation matrix of RAF, ERK1/2, MSK1 and CLOCK (Pearson’s r). Each point represents one biological replicate, color-coded by experimental group; solid lines show linear regression and shaded bands the 95% confidence interval. Spearman’s r and Pearson’s r with corresponding P values are given in each panel; n = 21 (3 independent replicates × 7 conditions)

Discussion

Overview of Principal Findings

In this study, we found that desflurane increased RAF–MAPK–MSK1 protein levels and reduced hippocampal CLOCK, and that these changes were associated with neuronal senescence, impaired proliferation, synaptic E/I imbalance, increased α-synuclein immunoreactivity, and cognitive impairment in aged C57BL/6 mice. Pharmacological inhibition of RAF with sorafenib attenuated these molecular and cellular changes and improved hippocampus-dependent learning and memory. By integrating in vitro mechanistic dissection in HT22 neurons with in vivo functional readouts spanning Western blot, immunofluorescence, whole-cell patch-clamp electrophysiology, and three independent behavioral paradigms, this work identifies RAF as a previously underexplored node that links kinase signaling, circadian, and senescence machinery to desflurane-induced cognitive impairment. The proposed mechanism is summarized schematically in Fig. 9.

Fig. 9.

Proposed mechanism linking desflurane exposure to reduced hippocampal CLOCK expression and cognitive impairment in aged mice. Desflurane exposure (2%, 2 h) increases hippocampal c-Raf, ERK1/2 and MSK1 protein abundance and reduces CLOCK expression in aged mice and HT22 hippocampal neurons. This is associated with cellular senescence, impaired proliferation, increased α-synuclein immunoreactivity and synaptic excitation–inhibition imbalance, and with impaired contextual and spatial memory. Genetic (siRNA knockdown, plasmid overexpression) and pharmacological (sorafenib) manipulation of RAF bidirectionally alters CLOCK levels, and RAF inhibition attenuates the associated cellular and behavioral changes

RAF–MAPK–MSK1 as an Upstream Regulator of Hippocampal CLOCK

The first key contribution of this work is evidence for a functional link between RAF–MAPK–MSK1 signaling (measured as protein abundance) and reduced CLOCK levels in hippocampal neurons (Figs. 1 and 4). Both genetic knockdown and plasmid-mediated overexpression of RAF produced bidirectional changes in CLOCK protein levels, consistent with RAF acting as an upstream regulator of CLOCK at the protein level. Because RAF-directed siRNA does not target the receptor tyrosine kinases or MEK5–ERK5 that sorafenib also inhibits, this genetic result argues against a purely off-target explanation in vitro. This finding is consistent with CLOCK acting downstream of RAF in hippocampal neurons. MSK1 phosphorylates CREB at Ser133 and histone H3 at Ser10/Ser28 to regulate activity-dependent gene transcription [33, 34], suggesting that elevated ERK1/2–MSK1 protein abundance in aged neurons may be associated with chromatin-level repression of CLOCK transcription or destabilization of CLOCK protein turnover. Although prior studies have repeatedly implicated BMAL1 deficiency in cognitive aging [23, 25], the contribution of its obligate heterodimeric partner CLOCK in the context of anesthetic-induced cognitive injury has remained comparatively underexplored. Our data position CLOCK as a kinase-regulated convergence point, expanding the conceptual framework of POCD beyond inflammation-centric models [9, 35] toward an integrated kinase–circadian axis.

Coupling of RAF Signaling to Senescence and Cell-Cycle Disturbance

A second mechanistic insight concerns the coupling between RAF signaling and the senescence–proliferation imbalance characteristic of aged hippocampal tissue (Figs. 2 and 3). Desflurane exposure simultaneously increased senescence-associated β-galactosidase activity and decreased EdU incorporation and S-phase entry, and RAF inhibition reversed both phenotypes in a concordant manner. As detailed in Results 3.2, matched-control comparisons (Des + si-RAF vs. Des + si-NC; Des + oe-RAF vs. Des + oe-NC) isolated RAF-specific effects from transfection-induced stress (Fig. 3H). This bidirectional coupling is biologically meaningful because RAS–RAF–MEK–ERK signaling is a canonical inducer of oncogene-induced senescence in mitotically competent cells through sustained p16INK4a/p21 activation [27, 28]. In the central nervous system, where mature neurons are largely post-mitotic, aberrant cell-cycle re-entry constitutes a recognized prelude to senescence-like states and neurodegeneration; recent single-cell transcriptomic analyses have shown that excitatory neurons that re-engage S-phase transcripts in the aging brain preferentially adopt a senescent fate and accumulate disproportionately in Alzheimer’s disease and Lewy body dementia [20–22]. RAF overexpression further suppressed EdU incorporation and S-phase progression while augmenting β-galactosidase signal on the background of desflurane exposure. This suggests that elevated RAF abundance may itself promote a senescence-permissive transcriptional state. This is concordant with prior work showing that BMAL1 deficiency accelerates oxidative stress-driven cellular senescence [24, 26]. Our model extends this chain one step upstream: desflurane → increased RAF–MAPK–MSK1 protein levels → reduced CLOCK levels → impaired circadian/oxidative homeostasis → senescence. The order of these events was not established experimentally, and this sequence is therefore proposed rather than demonstrated. This proposed chain reconciles cancer biology and circadian biology observations within a single neurological context.

Synaptic E/I Imbalance and Network Dysfunction

The electrophysiological findings extend the mechanistic implications of RAF–CLOCK signaling from cellular phenotype to circuit function (Fig. 6). Desflurane simultaneously reduced miniature excitatory postsynaptic current (mEPSC) amplitude and frequency while elevating miniature inhibitory postsynaptic current (mIPSC) parameters in hippocampal CA1 pyramidal neurons, producing a clear shift of the synaptic E/I balance toward inhibition. RAF inhibition partially restored this balance. The reduction in mEPSC parameters implicates both postsynaptic receptor availability and presynaptic release probability, while elevated mIPSC parameters indicate facilitated GABAergic transmission. Recent evidence from sevoflurane studies similarly demonstrates reduced hippocampal glutamatergic synaptic transmission in aged mice [12, 30, 36]. Our results add a defined upstream regulator—RAF kinase—to this pathway. Because MSK1 is required for BDNF-driven CREB phosphorylation and basal excitatory synaptic transmission [34], and CLOCK regulates dendritic morphology and synaptic plasticity [37], the protein-level changes we observed in hippocampal lysates identify these kinases as plausible molecular correlates of the synaptic phenotype, although the direction of this association is not straightforward: because only total protein was measured, the increase in ERK1/2 and MSK1 abundance cannot be equated with increased catalytic output and may instead reflect a compensatory response. Reduced network oscillatory burst frequency 24 h after exposure, together with the synaptic E/I imbalance observed at the terminal time point after completion of behavioural testing, indicates network-scale dysfunction spanning both the early and the later post-exposure phase.

α-Synuclein Accumulation: A Dual-Hit Mechanism

Increased hippocampal α-syn immunoreactivity in desflurane-exposed aged mice (Fig. 5) extends the growing evidence implicating α-syn pathology in perioperative neurocognitive disorders. Prior studies showed that propofol-surgery induces α-syn oligomerization in aged rat hippocampus [10], and that A53T α-syn transgenic mice exhibit heightened postoperative neurocognitive vulnerability [11]. The novelty of our observation lies in linking increased α-syn immunoreactivity specifically to RAF–MAPK signaling: pharmacological RAF inhibition substantially reduced α-syn immunofluorescence intensity, implicating RAF–ERK1/2 signaling in either α-syn upregulation or impaired α-syn clearance. Mechanistically, MAPK family kinases are known to phosphorylate α-syn and tau at threonine–proline motifs, promoting aggregation [38, 39], while CLOCK/BMAL1-regulated autophagy is required for α-syn clearance [10]. Our data therefore identify a plausible double-hit mechanism whereby increased RAF–MAPK–MSK1 signaling may both promote α-syn aggregation-prone modifications and—via reduced CLOCK levels—impair autophagic disposal. This dual action provides a plausible explanation for the increased α-syn immunoreactivity observed and positions RAF inhibition as a candidate intervention that simultaneously addresses both arms of α-syn dyshomeostasis.

Behavioral Validation and Translational Implications

The behavioral data (Fig. 7) validate the cognitive relevance of these molecular and cellular abnormalities. RAF inhibition improved performance across three orthogonal paradigms assessing anesthetic responsiveness (LORR/RORR), hippocampus–amygdala-dependent associative memory (CFC), and hippocampus-dependent spatial memory (MWM), providing internally consistent validation. Compared with traditional POCD models that rely on tibial fracture or laparotomy under inhaled anesthetic [40], our anesthetic-only paradigm isolates the contribution of desflurane exposure per se, thereby allowing more direct attribution of the observed cognitive deficits to volatile anesthetic-induced neuronal stress rather than to surgical trauma. This methodological distinction is relevant given the recent reclassification of perioperative neurocognitive disorders, in which anesthesia-specific effects are distinguished from surgical contributions [4]. The magnitude of cognitive improvement afforded by RAF inhibition—approaching control-level performance in CFC and partial restoration in MWM—is comparable to that reported for AMPK/SIRT1 activation [9] and NLRP3 inhibition [8], supporting RAF as a candidate target that warrants further investigation. Emerging cerebrospinal fluid biomarkers for perioperative neurocognitive disorders [35] may enable preoperative stratification and modulation of RAF pathway activity in future clinical trials.

Novelty in the Context of Existing POCD Literature

Three aspects of our findings extend the existing POCD literature. First, prior mechanistic accounts have emphasized neuroinflammation, microglial priming, and gut–brain axis dysregulation [41, 42], with limited attention to intracellular kinase–circadian crosstalk; the link between RAF–MAPK–MSK1 protein abundance and CLOCK introduces a complementary hippocampal mechanism that may interact with these inflammatory cascades. Second, although the RAS–RAF–MEK–ERK pathway is well characterized in oncology [43], its repositioning toward POCD remains largely unexplored, and the dose used here (50 mg/kg) [44] supports short-term perioperative feasibility [45]. Third, mechanistic data on desflurane remain markedly less abundant than for sevoflurane or isoflurane [12, 36].

Limitations and Future Directions

Several limitations warrant consideration. First, this study measured total protein levels of c-Raf, ERK1/2 and MSK1 but did not assess phosphorylation status. Changes in total protein abundance do not by themselves establish catalytic activation of the kinase cascade. Second, the sub-anesthetic exposure paradigm (2%, 2 h) does not recapitulate clinical maintenance at 1.0–1.5 × MAC with surgical trauma, and sorafenib is a multi-kinase inhibitor that also targets VEGFR, PDGFR and MEK5/ERK5 [31], although the concordant RAF-directed siRNA results argue against a dominant off-target contribution in vitro; clinically relevant MAC under surgical inflammation, more selective RAF inhibitors, or RAF-conditional knockout models would strengthen mechanistic attribution. Third, the mechanism linking MSK1 to CLOCK remains undefined, disrupted diurnal rhythmicity was not directly assessed, and the proposed dual-hit α-syn mechanism requires verification by phospho-Ser129-α-syn and autophagic flux markers (LC3-II, p62). Fourth, both male and female aged mice were used, but the group sizes were not sufficient to permit analyses stratified by sex; sex-dependent differences in RAF–MAPK–MSK1–CLOCK signalling and in vulnerability to desflurane therefore remain untested and should be examined in adequately powered, sex-balanced studies. Despite these limitations, the convergent multi-modal evidence provides a coherent, hypothesis-generating framework for RAF-directed studies of anesthetic-induced cognitive impairment.

Conclusion

In conclusion, the present study describes an inverse relationship between RAF–MAPK–MSK1 protein abundance and CLOCK levels in the aged hippocampus after desflurane exposure. The findings link several phenotypes observed in this model—neuronal senescence, impaired proliferation, E/I imbalance, α-synuclein immunoreactivity, and cognitive deficits—to a common upstream kinase, although the causal architecture among these phenotypes was not resolved. That pharmacological RAF inhibition attenuated these changes nominates RAF as a candidate target that warrants further preclinical and translational evaluation for the prevention or mitigation of anesthetic-induced cognitive impairment, an outcome of substantial clinical relevance given the projected rise in geriatric surgical volumes globally.

Supplementary Information

Below is the link to the electronic supplementary material.

Author Contributions

X.D. and X.S. conceptualized and designed the study and revised the manuscript critically for important intellectual content. Y.J. performed the experiments, collected and analyzed the data, conducted the statistical analysis, and wrote the initial draft of the manuscript. J.B., C.C. and X.Z. assisted with data collection, behavioral testing, and figure preparation. All authors reviewed and approved the final manuscript.

Funding

This work was supported by the Science and Technology Program of the Joint Fund of Scientific Research for the Public Hospitals of Inner Mongolia Academy of Medical Sciences (Grant No. 2024GLLH0632); the Health Research Project of the China Society for Metals, Metallurgical Safety and Health Branch (Grant No. jkws202433); the Research Project of Aerospace Medical and Health Technology Group Co., Ltd. (Grant Nos. 2024YK10 and 2025YK17); the Natural Science Foundation of Inner Mongolia Autonomous Region (Grant Nos. 2024MS08058 and 2025QN08075); the Inner Mongolia Medical University Joint Project (Grant No. YKD2024LH011); the Baotou Municipal Health Commission Research Project (Grant No. wsjkkj2025080); and the 2026 Postgraduate Research Innovation Project of Baotou Medical College (Grant No. BYKYCX202602).

Data Availability

The datasets generated and/or analysed during the current study are not publicly available due to institutional data protection policies but are available from the corresponding author on reasonable request.

Declarations

Competing Interests

The authors declare no competing interests.

Ethical Approval

All animal procedures were approved by the Ethics Committee of the Third Clinical Medical College, Inner Mongolia Medical University (Approval No. 2024-MER-013) and complied with the Regulations on the Administration of Laboratory Animals and the Guidelines for the Care and Use of Laboratory Animals.