METTL3-mediated m⁶A modification regulates CDKN1A to attenuate chronic sleep deprivation-induced cognitive impairment and neuronal apoptosis in rats

Abstract

Chronic sleep deprivation (CSD) can induce cognitive impairment, but its molecular mechanism remains unclear. In this study, initial m⁶A RNA sequencing of the hippocampal CA3 region in CSD rats, coupled with differential gene expression analysis of the total RNA fraction, revealed downregulation of METTL3, which was consistent with impaired performance in the Morris Water Maze (MWM) and confirmed by qRT-PCR and Western blot. Further investigation showed that, in HT-22 cells, METTL3 knockdown exacerbated rapamycin-induced apoptosis. RNA sequencing of METTL3-knockdown cells identified gene modules and specific differentially expressed genes associated with METTL3 loss. Differential expression analysis revealed that CDKN1A was significantly upregulated following METTL3 knockdown. Methylated RNA immunoprecipitation followed by qPCR (MeRIP-qPCR) further showed that METTL3 knockdown reduced the m⁶A methylation level of CDKN1A mRNA. In vivo, METTL3 overexpression in CSD rats reduced CDKN1A levels, decreased neuronal apoptosis, improved spatial memory, and alleviated CA3 neuronal damage. In vitro, METTL3 knockdown upregulated CDKN1A and promoted apoptosis in HT-22 cells, while CDKN1A knockdown reversed this effect. Collectively, our results demonstrate that METTL3 downregulation promotes CSD-induced cognitive impairment by driving CDKN1A-dependent neuronal apoptosis, thereby identifying the METTL3/CDKN1A axis as a potential therapeutic target.

Introduction

Cognitive impairment refers to the decline in cognitive functions such as memory and attention [1]. Sleep deprivation exerts a detrimental impact on the hippocampus, a brain region responsible for memory and learning [2]. Chronic sleep deprivation (CSD) not only impairs cognitive functions but also increases the risk of developing cognitive disorders and mental health conditions [3]. Additionally, CSD is associated with various health problems such as metabolic disorders and cardiovascular diseases, which can lead to elevated morbidity and mortality rates [4]. Therefore, developing of novel diagnostic biomarkers and prognostic indicators for CSD is essential to improve cognitive impairment.

The hippocampus plays a crucial role in the formation of long-term memory, among which the CA3 subregion serves as the core for memory encoding, storage, and retrieval. A study by Ochai J et al. demonstrated that CSD leads to impaired performance of rats in spatial memory tasks, accompanied by a reduction in the number of pyramidal cells, neuronal degeneration, and decreased Nissl substance staining intensity in the hippocampal CA3 subregion [5]. Watson JF et al. investigated the cellular and microcircuit characteristics of the human hippocampal CA3 region and found that this subregion exhibits a structural feature of sparse connectivity, which optimizes memory storage capacity [6]. Together, these findings suggest that the function of neurons in the hippocampal CA3 subregion may help reveal the molecular mechanisms underlying cognitive impairment.

In light of the vulnerability of the hippocampus to environmental insults, recent attention has turned to epigenetic regulation as a key mechanism underlying cognitive impairment [7]. N6-methyladenosine (m⁶A) is the most abundant internal modification in eukaryotic mRNA and has become a key epitranscriptome marker in the central nervous system, which participates in neural development, synaptic plasticity, and learning-memory processes by regulating RNA metabolism [8–10]. A study by Yan L et al. indicated that exercise promotes the restoration of m⁶A methylation in the brain through liver metabolism, improves neural network activity, and alleviates anxiety [11]. Among m⁶A regulatory components, Methyltransferase 3 (METTL3), as a major methyltransferase, is a crucial enzyme responsible for the methylation of RNA, particularly m⁶A, which is essential in modulating RNA stability, splicing, and translation [12, 13]. Additionally, METTL3 is engaged in several brain processes, such as neurogenesis and synaptic plasticity [14], and may regulate neuronal function participating in the pathophysiological processes of cognitive impairment [15]. Silencing of METTL3 reduces the proliferation of neural stem cells and impairs neuronal maturation [14, 16]. Chen H et al. have demonstrated that in an Alzheimer disease mouse model, METTL3 alleviated cognitive impairment and mitochondrial dysfunction by upregulating mitofusin 2 (MFN2) expression through m⁶A modification [17]. Together, these studies have indicated that METTL3 is a vital part of modulating both neuronal dysfunction and neuroinflammation to protect against cognitive impairment in various disease models.

CSD is known to induce cognitive impairment, yet the underlying molecular mechanisms remain poorly understood. Increasing evidence suggests that epitranscriptomic regulation, particularly m⁶A modification, plays a critical role in neuronal function, stress responses, and memory formation. The methyltransferase METTL3 plays an important role in regulating cell apoptosis, cell cycle, and RNA stability [18]. Deficiency of METTL3 prolongs the expression of pro-apoptotic genes, leading to abnormal apoptosis of cerebellar granule cells [19]. Under oxidative stress, METTL3 participates in the regulation of cellular antioxidant response and apoptosis by modulating m⁶A methylation [20]. NSUN2 and METTL3 can upregulate p21 expression through m⁶A/m⁵C methylation, thereby inhibiting hydrogen peroxide-induced cell apoptosis [21]. It is known that METTL3 downregulation is associated with cognitive impairment, but the causal relationship between CSD and METTL3 dysregulation—as well as the downstream mechanisms linking METTL3-dependent m⁶A modification to neuronal apoptosis—remains largely unclear. To address this knowledge gap, this study aims to elucidate how METTL3 contributes to CSD-induced cognitive deficits, with a particular focus on m⁶A-dependent regulatory pathways and neuronal apoptosis. Through in vivo and in vitro models, potential therapeutic targets for alleviating CSD-induced cognitive decline will be explored.

Material and methods

CSD rats model

All animal experiments were conducted in accordance with the ARRIVE guidelines (Animal Research: Reporting of In Vivo Experiments) and the National Institutes of Health Guide for the Care and Use of Laboratory Animals. All procedures were approved by the Ethics Committee of the First Affiliated Hospital of Zhengzhou University (approval number: 2025-KY-0434-001). In this study, a total of fifty 5-week-old male SD rats were used, purchased from Wuhan Servicebio Technology Co., Ltd. (Wuhan, China). The animal study consisted of two experimental phases. Phase I included two groups (Control and CSD, n = 10 per group). Phase II included three groups (Control, CSD + vector, and CSD + over-METTL3, n = 10 per group) to evaluate the impact of METTL3 overexpression. The randomization was implemented using a computer-generated random number table: each rat was assigned a unique number, and numbers were randomly sorted into 5 groups with equal size, ensuring no subjective selection bias. The rats were kept in soundproof rooms under controlled conditions (temperature: 24 ± 1 °C, 12-h light/dark cycle, relative humidity: 50-60%, starting at 8:00 AM) and permitted free access to standard laboratory water and food. Before the induction of CSD, all animals underwent a 3-week adaptation period. To induce CSD, a gentle method involving a rotating rod placed above the floor of the cage was used. The rats were gently nudged awake by the slow rotation of the rod, which encouraged low-level activity until the animals were fully awake. This procedure effectively induces CSD and has been validated in previous studies using behavioral wakefulness monitoring [22, 23]. This procedure effectively induces CSD, as described by previous studies. The rotation was programmed to accelerate from 4 to 40 rpm over 5 min. Each cycle of rotation lasted 5 min and was applied every 30 min throughout the 10-h deprivation period (from 8:00 AM to 6:00 PM). The direction of rotation was alternated periodically to minimize habituation, and rat activity was monitored to ensure wakefulness during the entire deprivation period. The CSD group was subjected to 10 h of sleep deprivation daily (from 8:00 AM to 6:00 PM) for a total duration of 6 weeks. The sample size of 10 rats per group was determined based on previous studies investigating hippocampal cognitive function and neuronal apoptosis in rodent sleep deprivation models. Power analysis (using G*Power 3.1 software) confirmed that this sample size achieves a statistical power of >80% (α = 0.05) to detect significant differences in key outcomes between groups, ensuring the reliability of statistical results.

Viral vector preparation and stereotaxic injection

For groups receiving viral injections (CSD + vector and CSD + over-METTL3), stereotaxic injection of adenovirus into the bilateral hippocampal CA3 region was performed immediately after the 1-week adaptation period. Following viral injection, rats were allowed a 2-week recovery period to ensure stable expression of the transgenes, after which CSD was induced. Adenovirus vectors for METTL3 overexpression (pEZ‑AV12-METTL3/mCherry) and the corresponding negative control vectors (pEZ‑AV12-vector/mCherry) were purchased from GenePharma (GenePharma, Shanghai, China). Stereotaxic microinjection was performed as previously described [24]. Briefly, rats were anesthetized with 4–5% isoflurane and maintained with 1–2% isoflurane (RWD LifeScience Co., Shenzhen, China) to ensure the stereotaxic injection procedure was performed painlessly, in line with the principle of minimizing animal suffering during experimental operations and securely placed in a stereotaxic frame. A total volume of 1.0 µl of adenovirus suspension (0.5 µl per side, 2 × 10⁹ titer units/ml diluted 10 × with enhanced infection solution) was bilaterally injected into the hippocampus using the following coordinates relative to bregma: anteroposterior (AP) -3.80 mm, mediolateral (ML) ± 4.2 mm, and dorsoventral (DV) -3.80 mm from the skull surface. The injection was performed over 5 min to allow for slow diffusion of the viral solution, and the needle was left in place for an additional 5-minute post-injection to minimize backflow. Rats injected with empty adenovirus vectors served as the negative control group (vector). Following the procedure, rats were allowed to recover on a heating pad until fully awake and were monitored daily for body weight, neurological status, and any signs of distress or infection. To protect the injection site, rats were housed individually for 24 h post-surgery.

Inclusion criteria: healthy male SD rats without surgical complications and with normal pre-operative behavior. Exclusion criteria: before data analysis, animals exhibiting severe illness or signs of infection, those in which stereotaxic surgery was unsuccessful or could not be recovered, or those with technical issues (e.g., misplacement of the injection needle) were excluded, and these predefined exclusion criteria (e.g., >20% body weight loss, postoperative neurological deficits) were recorded to ensure experimental transparency and reproducibility.

Morris water maze (MWM) test

As previously mentioned, the MWM test was conducted [25]. The maze consisted of a circular pool with a depth of 40 cm and a diameter of 120 cm, which was divided into four equal quadrants. The temperature of the water was maintained at 22 ± 1 °C, and the water was made opaque with the addition of milk to obscure a concealed platform in the middle of one of the quadrants, 1 cm below the water’s surface. Rats underwent daily trials for 5 consecutive days for training. During each trial, animals had little more than sixty seconds to find the secret platform, and once found, they had fifteen seconds to remain on the platform. A rat was manually directed to the platform and left there for 15 seconds if it was unable to find it within the allotted 60 seconds. On Day 6, the hidden platform was removed, and an exploration test was conducted. The rats were given 60 seconds to explore the pool, and their activity was tracked using a video tracking system (Jiliang Software Technology, Shanghai, China). Spatial memory performance was quantified using: (1) Target-quadrant time: cumulative time spent in the quadrant where the platform was previously located. (2) Platform crossings: number of times the rat crossed directly over the previous platform location. These measures were used to evaluate learning and memory performance, and all data were analyzed across individual animals.

Hematoxylin-eosin (H&E) staining

After the designated experimental period, rats were deeply anesthetized with an overdose of sodium pentobarbital (100 mg/kg, intraperitoneally) and transcardially perfused with 0.9% saline, followed by 4% paraformaldehyde (PFA, P1110, Solarbio, China) in phosphate-buffered sa Sline (PBS, pH 7.4). Craniotomy was performed to extract whole brains, which were post-fixed in the same fixative at 4 °C for 24 h. Using stereotaxic coordinates from Paxinos and Watson rat brain atlas (2007), the hippocampal CA3 subregion was identified and microdissected. The brain tissues containing the hippocampal CA3 region of the rats were sectioned, paraffin-embedded, and fixed in 4% paraformaldehyde for 24 h. The stained hippocampal CA3 region sections were hydrated in a series of ethanol solutions (100%, 95%, 85%, and 70%) for 5 min each. Following this, the sections were given three five-minute washes in tap water. After 10 min of hematoxylin staining, the slices were rinsed with water, differentiated with 1% hydrochloric acid alcohol for 10 seconds, rinsed with water again, and then rinsed with running water after 0.6% ammonia became blue. After three min in the eosin dying solution, the slices were dehydrated with ethanol and xylene to make them transparent, then air-dried for neutral gum sealing. Finally, the stained hippocampal CA3 region sections were examined by an Olympus light microscope in Tokyo, Japan.

Nissl staining

The brain tissues containing the hippocampal CA3 region of the rats were hydrated in a series of ethanol solutions. Following this, three rounds of five-minute washings were performed on each section. The tissue was then stained with 0.1% cresyl violet (41830-80-2, Sigma-Aldrich, Germany) for 10 min. Following staining, distilled water was used to rinse the sections, followed by a gradient of alcohol solutions to dehydrate them and xylene to clarify them. Finally, the sections were mounted using neutral gum. The stained sections were imaged using an Olympus VS120 Virtual Slide Scanner. The damaged neurons in the CA3 area were seen to be sickle-shaped, undersized, and highly stained, sometimes with fragmented or nonexistent nuclei, in contrast to normal neurons that were uniformly shaped and stained. The percentage of injured neurons in the CA3 area was calculated to assess the impact of sleep deprivation on hippocampal neuronal structure. For quantification, the region of interest (ROI) was defined as the cell-dense neuronal layer of the hippocampal CA3 region in each section. Neuronal counting was performed using automated segmentation in ImageJ software (version 2.0.0). To reduce subjectivity, brightness and contrast thresholds were first set on pilot images and then fixed across all images within the same staining batch. Automated counts were validated against manual counts for a subset of images (n = 15), yielding high concordance (Pearson r > 0.9). Counting was performed by an investigator blinded to the experimental groups. Only neurons with clearly visible nuclei were included, while ambiguous or overlapping cells were excluded. The percentage of injured neurons within the defined ROI was calculated to assess the impact of sleep deprivation on hippocampal neuronal structure.

Cell line and culture conditions

The Chinese Academy of Sciences Cell Bank (Shanghai, China) provided the mouse hippocampus neuronal cell line HT-22 cells. A complete medium consisting of 90% Dulbecco’s Modified Eagle’s Medium (11995065, DMEM, Gibco, USA) and 10% fetal bovine serum (A5669701, FBS, Thermo Fisher scientific, USA) was employed to cultivate HT-22 cells. The cells were cultured in a humidified incubator at 37 °C with an atmosphere of 5% CO₂.

Cell transfection and treatment

HT-22 cells were planted at a density of 2 × 10⁵ cells per well in 24-well plates and cultured overnight in complete growth medium to achieve approximately 70-80% confluency. Following the manufacturer’s guidelines, Lipofectamine 2000 transfection reagent (11668019, Invitrogen, Shanghai, China) was utilized. Specifically, small interfering RNAs (siRNAs) targeting METTL3 (si-METTL3-1 and si-METTL3-2), targeting CDKN1A (si-CDKN1A), or a negative control siRNA (si-NC), as well as plasmids encoding CDKN1A for overexpression, were all purchased from GenePharma (Shanghai, China) and transfected into HT-22 cells. The control group was transfected with an empty vector (vector, GenePharma, Shanghai, China) under the same conditions. All transfection procedures followed a standardized protocol as per the manufacturer’s guidelines. Cells were harvested for subsequent experimentation after 48 h of transfection. Rapamycin (Rap, S1039, Selleck, Houston, USA) was dissolved in dimethyl sulfoxide (DMSO) to generate a 10 mM stock solution. Before experimental treatment, the stock solution was further diluted in a full culture medium to 50 µg/ml as the ultimate concentration. Rap was administered to HT-22 cells for four hours at a dosage of 50 µg/ml [26].

m⁶A RNA sequencing analysis of rat hippocampal CA3 tissues

Total RNA was extracted from rat hippocampal CA3 tissues, including both control and CSD groups, using TRIzol reagent (R110, Solarbio, Beijing, China) following the manufacturer’s instructions. The concentration and purity of the RNA were measured using a NanoDrop 2000 spectrophotometer (Thermo Scientific, Shanghai, China), and RNA integrity was assessed with an Agilent 2100 Bioanalyzer (Agilent, Shanghai, China). For m⁶A RNA sequencing, methylated RNA immunoprecipitation (MeRIP) was performed using the GenSeq™ m⁶A-MeRIP Kit (GenSeq, China) according to the manufacturer’s protocol to enrich m⁶A-modified RNA fragments. The enriched RNA fragments were then used to construct sequencing libraries. Library preparation and sequencing were carried out by Majorbio Bio-Pharm Technology Co., Ltd. (Shanghai, China). Sequencing was performed on the Illumina NovaSeq 6000 platform, generating high-throughput, high-quality data for downstream m⁶A peak identification and differential methylation analysis.

RNA sequencing analysis of HT-22 cells transfected with si-METTL3-1 plasmids

Total RNA was extracted from HT-22 cells transfected with si-METTL3-1 and si-NC (negative control) plasmids using TRIzol reagent, following the manufacturer’s instructions. RNA concentration, purity, and integrity were measured using NanoDrop 2000 and Agilent 2100 Bioanalyzer, respectively. RNA sequencing libraries were prepared using the VAHTS Universal V6 RNA-seq Library Prep Kit (Vazyme, China) according to the manufacturer’s protocol. Library construction and sequencing were performed by Majorbio Bio-Pharm Technology Co., Ltd. on the Illumina NovaSeq 6000 platform, providing high-quality transcriptome data for analyzing the effects of METTL3 knockdown.

Dataset source and differential expression analysis

This study generated two sequencing datasets: m⁶A RNA-seq and RNA-seq. The m⁶A RNA-seq dataset included three control (normal rat) samples and three CSD (CSD rat) samples. The RNA-seq dataset consisted of three samples treated with si-NC (control group) and three samples transfected with si-METTL3-1 (si-METTL3 group). Differential expression analysis of genes was conducted separately for both datasets using the “limma” package in R software (version 4.1.0, https://www.r-project.org/). Genes were selected based on fold change (FC), with upregulated differentially expressed genes (DEGs) defined as those with FC > 2, and downregulated DEGs defined as those with FC < 0.5. Statistical significance was determined using the P < 0.05 criterion.

Hierarchical clustering analysis

Hierarchical clustering analysis was performed to assess the similarity among the 92 DEGs in the m⁶A RNA-seq dataset. The clustering results were visualized as a dendrogram, representing the pairwise relationships between arrays. The similarity between two arrays was measured by their correlation coefficient, with the association inversely related to the distance between the arrays. The clustering algorithm identified pairs of arrays with the smallest distance and linked them into clusters, with the distance between arrays defined as one minus their correlation. The algorithm iteratively built the tree structure by progressively merging the closest clusters, resulting in a binary hierarchical tree.

Weighted gene co-expression network analysis (WGCNA)

Weighted Gene Co-expression Network Analysis (WGCNA) was performed on the RNA-seq data to identify co-expressed gene modules. The co-expression network for all genes was constructed using the Bioinfo Intelligent Cloud platform (https://www.bic.ac.cn/BIC/#/). A scale-free network was built using a soft threshold with β = 18. Genes were clustered into several modules based on their co-expression patterns, with different colors representing each module. The two most strongly correlated modules were selected for further analysis. We detected module membership (MM) and gene significance (GS) to assess the stability of the identified modules.

Gene overlapping identification

To identify overlapping genes between DEGs from the RNA-seq data and the two most strongly correlated modules identified through WGCNA, we conducted a topological analysis using the Venn Online Graph Tool (https://bioinformatics.psb.ugent.be/webtools/Venn/).

Functional enrichment analysis

Functional enrichment analysis was conducted using Enrichr (https://maayanlab.cloud/Enrichr/) to investigate the biological relevance of DEGs identified from m⁶A RNA-seq data of CSD rat models and controls. This comprehensive analysis included Kyoto Encyclopedia of Genes and Genomes (KEGG) pathway and biological process (BP) enrichment. The GO classification system classifies gene and protein functions into three major hierarchical categories: BP, molecular function (MF), and cellular component (CC). In addition, GO (BP, CC, MF) and KEGG pathway enrichment analyses were conducted on DEGs identified from RNA sequencing data of HT-22 cells transfected with si-NC and si-METTL3-1.

Expression analysis and visualization

Expression levels of METTL3 in the m⁶A RNA-seq dataset, as well as the expression of overlapping genes in the RNA-seq data from HT-22 cells transfected with si-NC and si-METTL3-1, were analyzed using the Sangerbox platform (version 3.0, http://vip.sangerbox.com/home.html). First, the expression of METTL3 was assessed in both the control and CSD groups from the m⁶A RNA sequencing data. Subsequently, the expression of overlapping genes was analyzed between the si-NC and si-METTL3-1 groups within the RNA-seq dataset. When P < 0.05, the results obtained are significant.

Quantitative real-time PCR (qRT-PCR) analysis

Following the manufacturer’s instructions, total RNA was extracted from rat hippocampal tissue and mouse neurons with TRIzol reagent (R110, Solarbio, Beijing, China). The concentration and purity of RNA were assessed employing a NanoDrop 2000 spectrophotometer (Thermo Scientific, Shanghai, China). Following the manufacturer’s instructions, the process transformed 1 µg of total RNA into complementary DNA (cDNA) with the PrimeScript RT Master Mix Kit (RR036A, Takara, Japan). By a StepOnePlus Real-Time PCR System (Applied Biosystems, USA) and SYBR Green PCR Master Mix (639676, Takara, Dalian, China), qRT-PCR was carried out. The relative gene expression was calculated using the 2^-ΔΔCT method [27]. For rat hippocampal tissue samples, the following primers were used: METTL3 forward: 5’-ATCCCCAAGGCTTCAACCAG-3’, METTL3 reverse: 5’-ATCCAGTTGGGCTGCACATT-3’ and GAPDH forward: 5’-ACGGGAAACCCATCACCATC-3’, GAPDH reverse: 5’-TCACAAACATGGGGGCATCA-3’. For HT-22 cells, the following primers were used: METTL3 forward: 5’-GTCTTGCCATCTCTACGCCA-3’, METTL3 reverse: 5’-TCATGGCAGACAGCTTGGAG-3’, CDKN1A forward: 5’-TAAGGACGTCCCACTTTGCC-3’, CDKN1A reverse: 5’-AAAGTTCCACCGTTCTCGGG-3’, and GAPDH forward: 5’-GGGTCCCAGCTTAGGTTCAT-3’, GAPDH reverse: 5’-TGAGGTCAATGAAGGGGTCG-3’. The GAPDH gene was used as the internal reference for both rat and mouse samples.

GAPDH was used as the internal reference for normalization. Primer efficiency (90–110%) and linearity were confirmed using standard curves, and melting curve analysis verified product specificity. Each reaction was run in triplicate, and statistical analyses were based on biological replicates. Only data within the linear amplification range were used for quantification.

Western blot (WB) analysis

Protein lysates were prepared from rat hippocampal tissue and mouse neurons utilizing RIPA lysis buffer (#P0013B, Beyotime, China) with protease and phosphatase inhibitors added. The BCA Protein Assay Kit (Bioteke, Beijing, China) was applied to detect the protein concentration. After being separated by SDS-PAGE, equal quantities of protein were transferred to PVDF membranes (#P0965-20pcs, Beyotime, China). The membranes were then incubated with primary antibodies: anti-METTL3 (1:1000, ab195352, Abcam, Cambridge, UK), anti-CDKN1A (1:1000, ab109199, Abcam, Cambridge, UK), anti-Bax (1:1000, ab32503, Abcam, Cambridge, UK), anti-Caspase3 (1:2000, ab184787, Abcam, Cambridge, UK), anti-Cleaved Caspase3 (1:2000, ab184787, Abcam, Cambridge, UK) and anti-GAPDH (1:10000, ab181602, Abcam, Cambridge, UK), all of which were purchased from Abcam, Cambridge, UK. After incubation with primary antibodies, the membranes were incubated with Goat Anti-Rabbit IgG H&L (HRP, 1:2000, ab6721, Abcam, Cambridge, UK) secondary antibody. A technique for ECL detection (#P0018S, Beyotime, Shanghai, China) was utilized to observe protein bands. Images were captured and analyzed using ImageJ software (version 2.0.0).

GAPDH was used as the loading control to ensure equal protein loading and normalization across samples. Exposure times were optimized to maintain signals within the non-saturating range, and each blot was confirmed to show linear signal intensity with increasing protein load. Band intensities were quantified using ImageJ after background subtraction. All analyses were based on at least three independent biological replicates, with technical duplicates for each sample to ensure reproducibility.

Flow cytometry analysis

Flow cytometry was used to detect apoptosis in HT-22 cells. Initially, HT-22 cells were planted at a density of 1 × 10⁴ cells per well in 24-well plates. After dissociation with 0.25% trypsin-EDTA (25200056, Thermo Fisher Scientific, Inc.) and washing with PBS (C0221A, Beyotime, China), cells were incubated with 5 μl of PI solution and 5 μl of Annexin V-FITC for 15 min at room temperature to assess apoptosis. Following staining, a flow cytometer (ACEA NovoCyte 2060 R, China) was used to detect the fluorescence of Annexin V-FITC and PI, and FlowJo software (version 10.6.0, FlowJo software, Oregon, USA) was applied for data analysis.

m⁶A-RNA Immunoprecipitation followed by qRT-PCR (MeRIP-qPCR)

To assess the impact of METTL3 knockdown on m⁶A modification of CDKN1A mRNA, m⁶A-RIP followed by qRT-PCR was performed. Total RNA was extracted using TRIzol reagent and treated with DNase I to remove genomic DNA contamination. RNA was fragmented according to the mRNA Fragmentation Protocol provided by New England Biolabs (NEB, Beijing, China), typically by incubating at 94 °C for 5 min in fragmentation buffer. Fragmented RNAs were subjected to immunoprecipitation with m⁶A-specific antibody (Active Motif, Beijing, China) or IgG antibody (1:1000, ab172730, Abcam, Cambridge, UK) using the EpiMark N6-Methyladenosine Enrichment Kit (E1610S, New England Biolabs, USA), following the manufacturer’s instructions. After immunoprecipitation, RNA was purified and reverse-transcribed, followed by qPCR amplification. GAPDH mRNA was analyzed in parallel as a non-enriched internal control. m⁶A enrichment was calculated, normalized to input RNA and IgG control.

RNA stability assay

To evaluate the stability of CDKN1A mRNA, actinomycin D (SBR00013, Sigma-Aldrich, Germany) was used following METTL3 knockdown in HT-22 cells. As previously indicated, the test was performed. Actinomycin D was administered at a final 2 μg/mL concentration when the cells achieved 80% confluence. Following treatment, cells were then taken at 0, 2, 4, 6, and 8 h. Total RNA was extracted from the cells, and RNA stability was assessed by qRT-PCR.

Statistical analysis

The GraphPad Prism software (version 8.1.0) was applied to assess the data. Each experiment was conducted three times, and all data are presented as the mean ± standard error of the mean (SEM). Data were tested for normality using the Shapiro–Wilk test and for homogeneity of variance using Levene’s test. The differences between two groups were examined with the Student’s t-test. For multiple group comparisons, after testing for homogeneity of variance using Levene’s test as mentioned above, one-way analysis of variance (ANOVA) with Tukey’s or Bonferroni’s post hoc test was utilized for multiple group comparisons. For data involving repeated measurements over time within the same samples, repeated measures ANOVA was performed to assess statistical differences. The assumption of sphericity was evaluated using Mauchly’s test, and when violated (p < 0.05), the Greenhouse–Geisser correction was applied to adjust the degrees of freedom. P < 0.05 was regarded as statistically significant.

Results

METTL3 is significantly down-regulated in the hippocampal CA3 region of CSD rats

To investigate m⁶A methylation and concurrent gene expression changes in CSD rats, we performed m⁶A RNA-seq on rat hippocampal CA3 tissues. We evaluated m⁶A RNA-seq data quality via genome region distribution and gene body coverage analyses. The reads distribution histogram (Supplementary Figure 1A) showed all samples (control: 1-3, CSD: 1-3) had > 60% of reads enriched in CDS, with low proportions in other regions (e.g., TSS flanking regions, UTR, intergenic regions, introns), indicating high transcript capture specificity. The gene body coverage curve (Supplementary Figure 1B) confirmed consistent patterns across samples: coverage peaked in the middle gene body (≈20-60 th percentile, 5’ → 3’) and gradually declined toward the 3’ end, consistent with typical RNA-seq characteristics, verifying data reliability and reproducibility. The m⁶A RNA-seq analysis of rat hippocampal CA3 tissues revealed 43 significantly downregulated and 49 upregulated DEGs between control and CSD groups (Fig. 1A), with METTL3 exhibiting marked downregulation (indicated in red). Hierarchical clustering analysis of gene expression profiles further confirmed the obvious differences between the CSD group (n = 3) and the control group (n = 3) (Fig. 1B). GO analysis was then conducted to identify enriched biological processes (BP), and the enriched items included “Negative regulation of cell cycle (GO: 0045786)”, “Regulation of hormone secretion (GO: 0046883)”, “Mitotic G2/M transition checkpoint (GO: 0044818)”, etc (Fig. 1C). These processes indicated that CSD exposure may influence cell cycle progression and differentiation. KEGG pathway analysis revealed 5 significantly enriched pathways, including “Phenylalanine, tyrosine and tryptophan biosynthesis”, “Phenylalanine metabolism”, “Oxytocin signaling pathway”, and “Melanoma” (Fig. 1D). Finally, the expression level of METTL3 was analyzed, and the findings indicated that METTL3 expression in the CSD group markedly declined in comparison to the control group, suggesting that it might be involved in the CSD response (Fig. 1E).

Fig. 1. Differential gene expression and functional analysis in the hippocampal CA3 region of CSD rats.

A Volcano plot showing the differential expression of genes in m⁶A RNA-Seq data. Each point represents a gene, with the x-axis indicating the log₂ fold change and the y-axis indicating the -log₁₀ p-value. Genes that are significantly upregulated (orange) or downregulated (blue) are labeled, with METTL3 (highlighted in red) showing significant expression differences. B Heatmap of the DEGs between control (n = 3) and CSD groups (n = 3). The color scale represents the normalized expression levels, with orange indicating higher expression and purple indicating lower expression. Samples are clustered by group, with the control samples in green and the CSD samples in red. C GO enrichment analysis of DEGs. The x-axis represents the gene ratio, and the y-axis shows the top enriched biological processes. The size of the circles corresponds to the number of genes involved, while the color intensity represents the significance (-log₁₀ p-value). D KEGG pathway analysis showing the top enriched pathways associated with the DEGs. The x-axis represents the gene ratio, and the y-axis lists the pathways. Circle size represents the number of DEGs involved in each pathway, while color intensity indicates the significance of each pathway (-log₁₀ p-value). E Boxplot showing the expression levels of METTL3 in the CSD and control groups. The y-axis represents METTL3 expression levels. *P < 0.05 vs. Control group. CSD: Chronic sleep deprivation, DEGs: Differentially expressed genes, GO: Gene ontology, KEGG: Kyoto encyclopedia of genes and genomes.

METTL3 is downregulated in the hippocampal CA3 region in the CSD-induced cognitive impairment

Following the 6-week CSD modeling period, MWM tests were conducted immediately to assess cognitive function. A schematic timeline of the experimental procedure is presented in Fig. 2A. The trajectory patterns for both the control and CSD groups are shown in Fig. 2B. While the control group exhibited a typical path toward the target platform, the CSD group showed a more erratic trajectory. The frequency of crossing the platform (Fig. 2C) was substantially lower in the CSD group than in the control group, indicating impaired cognitive performance in CSD-exposed animals. In addition, the CSD group spent a notably less amount of time in the target quadrant (Fig. 2D) during the MWM test, further confirming the cognitive deficits induced by CSD exposure. Notably, differential gene expression analysis based on total RNA signals from m⁶A RNA-seq identified METTL3 as exhibiting significant downregulation in the hippocampal CA3 region of CSD rats. Moreover, METTL3 has been extensively studied in the context of hippocampal function and related neurological processes. To investigate the molecular changes underlying these cognitive impairments, we measured the relative mRNA expression of METTL3. As shown in Fig. 2E, METTL3 mRNA expression was significantly lower in the CSD group compared to the control group, indicating that METTL3 might contribute to CSD-induced cognitive dysfunction. WB analysis also showed a similar pattern, with the CSD group’s METTL3 protein expression significantly lower than that of the control group (Fig. 2F, G). These findings highlight a potential link between METTL3 downregulation and impaired cognitive function following CSD exposure.

Fig. 2. METTL3 is downregulated in the hippocampal CA3 region in the CSD-induced cognitive impairment.

A A schematic diagram illustrating the experimental timeline. Five-week-old male SD rats were subjected to a 21-day adaptation period, followed by 42 days of CSD induction. Afterward, MWM training was performed for 5 consecutive days, and testing was conducted on Day 6. B-D MWM behavioral tests. (B) Representative swimming path of the control and CSD groups in the MWM test (n = 10). The red dot indicates the starting point, and the green dot represents the target location. C Frequency of crossing the platform (n = 10). *P < 0.05 vs. Control group. D Time spent in the target quadrant during the MWM test (n = 10). *P < 0.05 vs. Control group. (E) Relative mRNA expression levels of METTL3 were significantly decreased in the CSD group compared to the control group, as determined by qRT-PCR (n = 10). *P < 0.05 vs. Control group. F, G WB analysis of METTL3 protein levels in the CA3 region and quantification of relative protein expression levels (n = 3). *P < 0.05 vs. Control group. CSD: Chronic sleep deprivation, qRT-PCR: Quantitative real-time reverse transcription PCR, WB: Western blot, MWM: Morris water maze.

METTL3 knockdown increases Rap-induced apoptosis in HT-22 cells

To explore the function of METTL3 in neurons, METTL3-specific siRNAs were transfected in HT-22 cells, and the effective knockdown of METTL3 was confirmed by qRT-PCR and WB experiments (Fig. 3A-C). Among the si-METTL3 plasmids, si-METTL3-1 exhibited higher knockdown efficiency, and therefore, si-METTL3-1 was chosen for further research. To determine the optimal concentration of rapamycin for subsequent experiments, CCK8 assay was performed to detect the cell viability of HT-22 cells treated with different concentrations of rapamycin. The results showed that cell viability decreased in a concentration-dependent manner with the increase of rapamycin concentration. When the concentration of rapamycin reached 50 μg/mL, the cell viability was approximately 50%, which was selected as the effective intervention concentration for subsequent experiments (Fig. 3D). To evaluate the impact of METTL3 knockdown on apoptosis, HT-22 cells were treated with Rap by flow cytometry analysis. The outcomes demonstrated that the apoptosis rate increased after Rap treatment alone, and the apoptosis rate was further enhanced after the knockdown of METTL3 on this basis, suggesting that METTL3 knockdown could enhance Rap-induced apoptosis (Fig. 3E, F). In addition, WB analysis identified Caspase-3, cleaved Caspase-3 and Bax expression, three proteins linked to apoptosis (Fig. 3G-I). The outcomes proved that Bax and cleaved Caspase-3 in the Rap+si-METTL3-1 group were considerably raised, confirming the function of METTL3 in regulating apoptosis.

Fig. 3. METTL3 knockdown aggravates Rap-induced apoptosis in HT-22 cells.

A qRT-PCR analysis showed that METTL3 mRNA expression was significantly reduced after siRNA-mediated knockdown (si-METTL3-1 and si-METTL3-2) in HT-22 cells (n = 3). *P < 0.05 vs. Control group. B, C WB and quantification confirmed decreased METTL3 protein expression following siRNA transfection (n = 3). *P < 0.05 vs. Control group. D CCK-8 analysis for the effect of different concentrations of rapamycin on cell viability (n = 3). E, F Flow cytometric analysis of apoptosis in HT-22 cells treated with Rap and/or si-METTL3 (n = 3). Early and late apoptotic populations were defined based on Annexin V–PI quadrant gating: Q1 (PI⁺/Annexin V⁻): necrotic cells, Q2 (PI⁺/Annexin V⁺): late apoptotic cells, Q3 (PI⁻/Annexin V⁺): early apoptotic cells, Q4 (PI⁻/Annexin V⁻): viable cells. *P < 0.05 vs. Control group; ^#P < 0.05 vs. Rap+si-NC group. G-I WB was used to detect the expression of apoptosis-related proteins in HT-22 cells before and after METTL3 knockdown and Rap induction (n = 3). *P < 0.05 vs. Control group; ^#P < 0.05 vs. Rap+si-NC group. CSD: Chronic sleep deprivation, qRT-PCR: Quantitative real-time reverse transcription PCR, WB: Western blot, CCK8: Cell Counting Kit-8, Rap: Rapamycin.

WGCNA identifies key gene co-expression modules associated with METTL3 knockdown in HT-22 cells

We performed WGCNA on RNA-seq data of METTL3 knockdown in HT-22 cells. Sample clustering showed that samples were clustered into different groups according to treatment, and no significant outliers were detected (Fig. 4A). 18 was found to be the network’s optimal soft threshold, and the scale-free topological fit index was 0.85 (Fig. 4B). The dynamic tree cutting and resulting module colors were allocated to different gene co-expression modules (Fig. 4C). These modules were then analyzed for their relationship with METTL3 knockdown (Fig. 4D). The eigenvalue adjacency heatmap demonstrated the correlation between modules and revealed strong adjacency within the modules, especially the MEbrown and MEgreen modules (Fig. 4E). Of these, the correlation coefficient between module members and gene importance (si-NC) of MEbrown was 0.99 (Fig. 4F). The correlation coefficient between module members and gene importance (si-METTL3-1) of MEgreen module was 0.93 (Fig. 4G). These module-specific correlations suggest that the MEbrown and MEgreen modules may represent gene clusters functionally involved in METTL3-associated molecular responses. Further enrichment analysis of these modules may help to identify pathways related to m⁶A-mediated neuronal regulation or stress responses.

Fig. 4. WGCNA analysis of RNA-seq data of METTL3-knockdown in HT-22 neuronal cells.

A Sample dendrogram with/without trait heatmap. The hierarchical clustering of samples based on gene expression shows clustering of siRNA and NC groups (n = 3). The heatmap beneath the dendrogram shows the trait distribution for each sample. B Selection of the soft threshold power for WGCNA. The plot shows the relationship between the soft threshold power (x-axis) and the scale-free topology fit index (y-axis). C Cluster dendrogram of genes. The hierarchical clustering of genes is shown, with gene modules assigned different colors according to the dynamic tree cut method. The colors at the bottom represent the different modules identified in the network. D EigenGene adjacency heatmap. The heatmap displays the correlation between ME from different modules. Strong positive correlations (red) and weak or negative correlations (blue) are indicated. E Correlation heatmap of module and gene significance. F, G Scatter plots of module membership versus gene significance in the brown (F) and green (G) modules. WGCNA: Weighted gene co-expression network analysis, ME: Module eigengenes.

Differential gene expression and pathway analysis in METTL3 knockdown cells

Differential gene expression analysis was conducted after knockdown of METTL3 in HT-22 cells. 171 upregulated DEGs and 50 downregulated DEGs were obtained, of which 14,634 genes did not change significantly (Fig. 5A). A total of nine overlapping genes between the DEGs and WGCNA modules were identified, with seven derived from the green module and two from the brown module. (Fig. 5B). GO analysis revealed that the pathways enriched in these overlapping genes included “DNA damage response (GO:0006974)”, “Mitotic G1 DNA damage checkpoint signaling (GO:0031571)”, DNA damage response (GO:0006974)”, “Mitotic G1 DNA damage checkpoint signaling (GO:0031571)”, and “Lysophosphatidic acid acyltransferase activity (GO:0042171)”, etc (Fig. 5C). These outcomes indicated that METTL3 knockdown may affect DNA damage response and cell cycle regulation. Additionally, KEGG pathway analysis indicated a significant enrichment in the “p53 signaling pathway” (Fig. 5D), further supporting the role of METTL3 in cell cycle and apoptosis-related pathways. Finally, we validated the expression levels of these nine overlapping genes, including PRELP, BTG2, ANXA8, CDKN1A, CDSN, INKA2, PIDD1, CES2E, and CRLS1. As shown in Fig. 5E, PRELP, BTG2, ANXA8, CDKN1A, INKA2, PIDD1, CES2E were upregulated after METTL3 knockdown, while CDSN and CRLS1 were downregulated. Among these, CDKN1A exhibited a more pronounced expression between the two groups and was therefore selected for subsequent experimental validation. These results further confirmed the molecular effects of METTL3 knockdown on gene expression.

Fig. 5. Differential gene expression and functional enrichment analysis in METTL3 knockdown HT-22 cells.

A Volcano plot depicting the differential gene expression between si-METTL3-1 group (n = 3) and NC groups (n = 3). Genes that are significantly upregulated (orange) or downregulated (purple) are highlighted. The x-axis represents the log₂ fold change, and the y-axis represents the -log₁₀ p-value. B Venn diagram showing the overlap between DEGs and the genes in the green and brown modules identified by WGCNA. C GO analysis of DEGs. The x-axis represents the gene ratio (proportion of DEGs annotated to each term), and the y-axis shows the terms. The size of the circles corresponds to the number of DEGs in each pathway, and the color intensity represents the significance of each pathway (-log₁₀ P-value). D KEGG pathway enrichment analysis of DEGs in the si-METTL3 group. The x-axis represents the gene ratio, and the y-axis lists the top enriched pathways. The size of the circles indicates the number of DEGs involved in each pathway, and color intensity represents the significance (-log₁₀ P-value). E Boxplot showing the expression levels of selected DEGs in METTL3 knockdown and NC groups. *P < 0.05 or **P < 0.01 or ***P < 0.001 vs. si-NC group. NC: Negative control, WGCNA: Weighted gene co-expression network analysis, GO: Gene ontology, KEGG: Kyoto encyclopedia of genes and genomes, DEGs: Differentially expressed genes.

Effect of METTL3 knockdown on CDKN1A expression and m⁶A modification in HT-22 cells

The impact of METTL3 knockdown on CDKN1A expression was examined by WB and qRT-PCR in HT-22 cells. In contrast with the si-NC group, both the mRNA and protein levels of CDKN1A were upregulated in the si-METTL3-1 group (Fig. 6A-C). Further analysis was conducted to determine the impact of METTL3 knockdown on the m⁶A modification of CDKN1A mRNA. The results revealed a substantial decrease in m⁶A enrichment on CDKN1A mRNA following METTL3 knockdown (Fig. 6D). Previous research has revealed that RNA m⁶A methylation plays a role in regulating RNA stability [28]. To assess the stability of CDKN1A mRNA, actinomycin D was used following METTL3 knockdown in HT-22 cells. The outcomes revealed that the si-METTL3-1 group’s CDKN1A mRNA stability was higher than that of the si-NC group (Fig. 6E). These findings imply that METTL3 regulates CDKN1A expression by modulating the m⁶A-mediated RNA degradation process.

Fig. 6. Effect of METTL3 knockdown on CDKN1A expression, m⁶A modification, and RNA stability in HT-22 cells.

A-C CDKN1A expression analysis after METTL3 knockdown. qRT-PCR (A) and WB (B, C) were performed to assess the mRNA and protein levels of CDKN1A in HT-22 cells transfected with si-NC or si-METTL3-1 (n = 3). *P < 0.05 vs. si-NC group. D Relative m⁶A enrichment fold change in HT-22 cells transfected with si-NC or si-METTL3-1 (n = 3). *P < 0.05 vs. si-NC group. E CDKN1A mRNA stability after METTL3 knockdown. Actinomycin D treatment was used to measure the stability of CDKN1A mRNA in HT-22 cells transfected with si-NC or si-METTL3-1 (n = 3). *P < 0.05 vs. si-NC group. qRT-PCR: Quantitative real-time polymerase chain reaction, WB: Western blot.

Impact of METTL3 overexpression on CDKN1A expression and apoptotic pathways in the hippocampal CA3 region of CSD rats

To investigate the effect of METTL3 overexpression in CSD-induced cognitive impairment rats, an adenovirus vector encoding METTL3 and the fluorescent reporter mCherry was injected into the hippocampal CA3 region. The mCherry fluorescence was examined to assess the accuracy of CA3 targeting and the efficiency of viral transduction. Robust red fluorescence signals were detected in the CA3 region, indicating successful viral delivery and effective METTL3 overexpression in the cells (Supplementary Figure 1C). To assess the function of METTL3 in regulating CDKN1A and hippocampal CA3 region tissue changes in CSD rats, WB analysis was applied to evaluate the levels of METTL3 and CDKN1A proteins in the hippocampal CA3 region following METTL3 overexpression. Investigations demonstrated that in the hippocampal CA3 region of CSD rats, CDKN1A expression increased whereas METTL3 expression decreased. However, upon METTL3 overexpression, METTL3 expression was partially restored, and CDKN1A levels significantly decreased (Fig. 7A-C). Subsequently, the levels of apoptosis-related proteins in the hippocampal CA3 region of CSD rats were evaluated. The results showed an increase in Bax and cleaved Caspase-3 levels in the hippocampal CA3 region, indicating that CSD may activate pro-apoptotic pathways, leading to neuronal apoptosis in the hippocampal CA3 region (Fig. 7D-F). Interestingly, after METTL3 overexpression, cleaved Caspase-3 and Bax levels were suppressed in the hippocampal CA3 region of CSD rats (Fig. 7G-I). These findings suggest that METTL3 overexpression modulates apoptosis-related pathways in the hippocampal CA3 region, potentially through the regulation of CDKN1A expression.

Fig. 7. METTL3 overexpression modulates CDKN1A expression and apoptotic pathways in the hippocampus of CSD rats.

A-C WB analysis of METTL3 and CDKN1A protein expression in the hippocampus of CSD rats following METTL3 overexpression. A Representative WB for METTL3 and CDKN1A proteins (n = 3). B Quantification of METTL3 expression (n = 3). C Quantification of CDKN1A expression (n = 3). *P < 0.05 vs. Control group; ^#P < 0.05 vs. CSD+vector group. D-F Expression of apoptosis-related proteins in the hippocampus of CSD rats. D Representative WB for Bax, Caspase-3 and cleaved Caspase-3 (n = 3). E Quantification of Bax expression (n = 3). F Quantification of cleaved Caspase-3 expression (n = 3). *P < 0.05 vs. Control group. G-I Effect of METTL3 overexpression on apoptosis-related protein expression in the hippocampus of CSD rats. G Representative WB for Bax, Caspase-3 and cleaved Caspase-3 (n = 3). H Quantification of Bax expression (n = 3). I Quantification of cleaved Caspase-3 expression (n = 3). *P < 0.05 vs. Control group; ^#P < 0.05 vs. CSD+vector group. WB: Western blot, CSD: Chronic sleep deprivation.

METTL3 overexpression mitigates cognitive dysfunction and hippocampal neuronal damage in CSD rats

To further evaluate the impact of CSD and METTL3 overexpression on cognitive function, the MWM test was used to evaluate the trajectory patterns of rats. The control group exhibited a typical, stable path toward the target platform, frequently crossing the platform and spending more time in the target quadrant. In contrast, the CSD + vector group showed fewer stable trajectories, with less time spent in the target quadrant and a lower number of platform crossings. Compared with CSD + vector group, the CSD + over-METTL3 group demonstrated more stable and direct paths toward the target platform. These rats traversed the platform more frequently and spent more time in the target quadrant, indicating that METTL3 overexpression could reverse the cognitive deficits induced by CSD (Fig. 8A-C). Next, the hippocampal CA3 region cells’ pathogenic alterations were observed by H&E staining. The results showed that the cells in the CA3 region of the hippocampus in the CSD + vector group were more severely atrophied in contrast to the control group. Conversely, the cells atrophy in the CA3 region of the CSD + over-METTL3 group was less pronounced, suggesting that METTL3 overexpression partially alleviates the neuronal damage caused by CSD (Fig. 8D). Nissl staining was used to assess the structural damage to hippocampal neurons in the CA3 region. In contrast to the control group, the CSD + vector group revealed a reduction in the number of Nissl-stained neurons, with irregular distribution. In contrast, the CSD + over-METTL3 group demonstrated an increase in the number of neurons in the CA3 region, with a more even distribution of Nissl bodies (Fig. 8E, F). These outcomes imply that METTL3 overexpression may restore neuronal damage induced by CSD, improving neuronal survival and function.

Fig. 8. Impact of CSD and METTL3 overexpression on cognitive behavioral and hippocampal neuronal damage in rats.

A MWM trajectory analysis showing the paths taken by rats from the control (n = 10), CSD + vector (n = 10), and CSD + over-METTL3 groups (n = 10). The red circle indicates the target area, and the tracks demonstrate differences in spatial navigation. B Quantification of platform crossings (n = 10). *P < 0.05 vs. Control group. ^#P < 0.05 vs. CSD+vector group. C Time spent in the target quadrant (n = 10). *P < 0.05 vs. Control group; ^#P < 0.05 vs. CSD+vector group. D Representative H&E staining images of hippocampal tissue to evaluate neuronal damage in the CA3 region (n = 3). Scale bars: 500 μm (top), 100 μm (bottom). E, F Representative Nissl staining images of hippocampal neurons in the CA3 region to assess structural damage and distribution. E Representative images of Nissl-stained neurons in the CA3 region (n = 3). The black box indicates the region of interest (ROI) used for quantification. F Quantification of injured cells/total cells in the enlarged region (%) based on Nissl staining (n = 3). Scale bars: 500 μm (top), 100 μm (bottom). *P < 0.05 vs. Control group; ^#P < 0.05 vs. CSD+vector group. CSD: Chronic sleep deprivation, MWM: Morris water maze, H&E: hematoxylin and eosin.

METTL3 knockdown promotes neuronal apoptosis through the CDKN1A dependent pathway in HT-22 cells

To explore the role of CDKN1A in hippocampal neurons, we first transfected HT-22 cells with a CDKN1A overexpression plasmid. Western blot and qRT-PCR analyses validated successful CDKN1A overexpression (Fig. 9A-C). We then examined whether METTL3 regulates CDKN1A expression. As shown in Fig. 9D-F, CDKN1A overexpression markedly elevated CDKN1A levels, and this effect was further enhanced by concurrent METTL3 knockdown, suggesting that METTL3 may suppress CDKN1A expression. Flow cytometry analysis revealed that CDKN1A overexpression alone promoted apoptosis, and METTL3 knockdown further exacerbated this pro-apoptotic effect (Fig. 9G, H). A control group (si-NC + over-CDKN1A) confirmed that the enhanced apoptosis was specifically due to METTL3 knockdown rather than non-specific siRNA effects.

Fig. 9. METTL3 knockdown promotes neuronal apoptosis through the CDKN1A dependent pathway in HT-22 cells.

A Confirmation of CDKN1A overexpression in HT-22 cells via qRT-PCR (n = 3). *P < 0.05 vs. Oe-NC group. B-C Confirmation of CDKN1A overexpression in HT-22 cells by WB. B Representative WB images showing CDKN1A protein levels. (C) Quantification of CDKN1A protein expression (n = 3). *P < 0.05 vs. Oe-NC group. D qRT-PCR analysis of CDKN1A mRNA expression in HT-22 cells following CDKN1A overexpression and/or METTL3 knockdown (n = 3). *P < 0.05 vs. Oe-NC group; ^#P < 0.05 vs. si-NC+over-CDKN1A group. (E-F) WB analysis of CDKN1A protein expression in HT-22 cells. E Representative WB images showing CDKN1A protein expression. F Quantification of CDKN1A protein expression in HT-22 cells under different treatment conditions (n = 3). *P < 0.05 vs. Oe-NC group; ^#P < 0.05 vs. si-NC+over-CDKN1A group. (G-H) Flow cytometry analysis of apoptosis in HT-22 cells following CDKN1A overexpression and/or METTL3 knockdown. G Representative flow cytometry plots showing apoptosis levels. Early and late apoptotic populations were defined based on Annexin V–PI quadrant gating: Q1 (PI⁺/Annexin V⁻): necrotic cells, Q2 (PI⁺/Annexin V⁺): late apoptotic cells, Q3 (PI⁻/Annexin V⁺): early apoptotic cells, Q4 (PI⁻/Annexin V⁻): viable cells. H Quantification of apoptosis percentage (n = 3). *P < 0.05 vs. Oe-NC group; ^#P < 0.05 vs. si-NC+over-CDKN1A group. I Validation of CDKN1A knockdown efficiency in HT-22 cells via qRT-PCR (n = 3). *P < 0.05 vs. si-NC group. J, K) Validation of CDKN1A knockdown efficiency in HT-22 cells by WB. (J) Representative WB images showing CDKN1A protein levels. K Quantification of CDKN1A protein expression (n = 3). *P < 0.05 vs. si-NC group. (L) Analysis of CDKN1A mRNA expression in HT-22 cells following co-knockdown of METTL3 and CDKN1A via qRT-PCR (n = 3). *P < 0.05 vs. si-NC group; ^#P < 0.05 vs. si-METTL3+si-NC group. (M–N) Analysis of CDKN1A protein expression in HT-22 cells following co-knockdown of METTL3 and CDKN1A by WB. M Representative WB images showing CDKN1A protein levels. N Quantification of CDKN1A protein expression (n = 3). *P < 0.05 vs. si-NC group; ^#P < 0.05 vs. si-METTL3+si-NC group. O, P Effect of co-knockdown of METTL3 and CDKN1A on cell apoptosis in HT-22 cells. O Representative flow cytometry plots showing apoptosis levels. Early and late apoptotic populations were defined based on Annexin V–PI quadrant gating: Q1 (PI⁺/Annexin V⁻): necrotic cells, Q2 (PI⁺/Annexin V⁺): late apoptotic cells, Q3 (PI⁻/Annexin V⁺): early apoptotic cells, Q4 (PI⁻/Annexin V⁻): viable cells. P Quantification of apoptosis percentage (n = 3). *P < 0.05 vs. si-NC group; ^#P < 0.05 vs. si-METTL3+si-NC group. WB: Western blot, qRT-PCR: quantitative reverse transcription PCR.

To definitively establish whether CDKN1A acts as a downstream mediator of METTL3 in neuronal apoptosis, CDKN1A knockdown and METTL3/CDKN1A co-knockdown experiments were performed. Transfection with si-CDKN1A effectively reduced CDKN1A expression at both mRNA and protein levels (Fig. 9I-K). We then assessed the consequences of co-knockdown. METTL3 knockdown alone (si-METTL3 + si-NC) significantly increased CDKN1A expression, whereas simultaneous knockdown of both METTL3 and CDKN1A (si-METTL3 + si-CDKN1A) reversed this upregulation, restoring CDKN1A to near-control levels (Fig. 9L-N). Consistent with these molecular changes, flow cytometry analysis demonstrated that the dramatic pro-apoptotic effect of METTL3 knockdown was reversed by concurrent CDKN1A knockdown (Fig. 9O-P). Together, these data indicate that METTL3 knockdown promotes neuronal apoptosis primarily by upregulating CDKN1A, and that CDKN1A is an essential downstream effector in this regulatory pathway.

Discussion

This study aimed to elucidate the mechanistic role of METTL3-mediated m⁶A RNA methylation in cognitive impairment induced by CSD. We demonstrated that CSD causes significant spatial learning and memory deficits in rats, accompanied by a marked downregulation of METTL3 expression in the hippocampal CA3 region. Mechanistically, METTL3 modulates m⁶A modification of CDKN1A mRNA in hippocampal neurons, thereby regulating its expression. Reduced METTL3 activity during CSD leads to increased CDKN1A levels, which in turn activate downstream pro-apoptotic factors Bax and cleaved caspase-3, ultimately triggering neuronal apoptosis. This cascade contributes to neuronal damage and the observed cognitive deficits (Fig. 10). Critically, METTL3 overexpression reversed these effects, restoring CDKN1A homeostasis, suppressing apoptosis, and rescuing cognitive deficits. This METTL3-CDKN1A axis highlights how sleep deprivation disrupts m⁶A-dependent RNA stability to amplify apoptotic signaling. Together, these results indicate that METTL3 exerts a neuroprotective effect in CSD-induced cognitive dysfunction, likely by modulating CDKN1A-mediated apoptotic pathways via m⁶A-dependent mechanisms.

Fig. 10. Molecular mechanism of CSD-induced neuronal apoptosis and cognitive dysfunction.

Schematic illustration of the proposed mechanism by which CSD induces neuronal apoptosis and cognitive dysfunction via the METTL3-m⁶A-CDKN1A axis. CSD: Chronic sleep deprivation, m⁶A: N6-methyladenosine, METTL3: methyltransferase-like 3, CDKN1A: cyclin-dependent kinase inhibitor 1A, Bax: Bcl-2-associated X protein, Caspase-3: cysteine aspartase-3.

Cognitive dysfunction, characterized by impairments in memory, perception, and thinking, arises from abnormalities in the cerebral cortex’s function and structure [29]. Numerous studies have highlighted a strong association between sleep deprivation and cognitive decline. For instance, Csipo T et al. have demonstrated that 24-h sleep deprivation significantly impaired cognitive performance, altered blood flow responses in the prefrontal and somatosensory cortices, and caused the right middle cerebral artery’s cerebral blood flow to decrease [30]. Additionally, research by Tang H et al. has indicated that the natural compound Osthole, through activation of the Nrf2/HO-1 pathway, notably improved cognitive performance in CSD rats, restored antioxidant enzyme activity, and reversed neuronal damage in the hippocampal CA1 region [31]. Similarly, Wang X et al. have found that melatonin alleviated cognitive deficits and hippocampal inflammation induced by sleep deprivation by modulating gut microbiota and metabolic products [32]. Our investigation created a CSD rat model and performed gene sequencing of hippocampal cells to explore underlying molecular mechanisms. Differential expression, hierarchical clustering, and enrichment analysis of the sequencing data identified METTL3 as a hub gene significantly downregulated in CSD rats. Behavioral testing using the MWM revealed pronounced cognitive deficits in CSD rats, accompanied by a reduction in METTL3 expression in their hippocampi. The observations imply that METTL3 could be essential in sleep deprivation-induced cognitive dysfunction and hippocampal alterations.

METTL3 plays a critical role in post-transcriptional RNA modifications, influencing gene expression within cells [33]. Several studies have suggested that dysregulated expression or dysfunction of METTL3 may contribute to cognitive impairments, particularly those associated with neurodevelopmental, neuroinflammatory, and neurodegenerative disorders [34, 35]. METTL3 forms a complex with METTL14, which catalyzes the addition of N6-methyladenosine (m⁶A) modifications to RNA molecules, thereby mediating the majority of m⁶A methylation on mRNA transcripts. For example, Huang H et al. have reported aberrant m⁶A methyltransferase METTL3 and RBM15B expression in the hippocampus of Alzheimer’s disease (AD) patients [36]. METTL3 accumulation in the insoluble fraction was found to correlate with insoluble Tau protein levels, suggesting that abnormal METTL3 expression and distribution could contribute to AD pathology through epitranscriptomic mechanisms. Similarly, He B et al. have demonstrated that the anesthetic isoflurane reduced METTL3 phosphorylation in the hippocampus by inhibiting the MAPK/ERK pathway, disrupting m⁶A RNA methylation signals, and potentially affecting postoperative cognitive dysfunction (POCD) [37]. Moreover, Ming Y et al. have identified that METTL3 stabilizes MALAT1 expression by promoting m⁶A modification in a mouse model of autism, which activates the Wnt/β-catenin signaling pathway and alleviates autism-like behaviors while inhibiting hippocampal neuronal apoptosis [38]. In our study, rapamycin was used to induce neuronal injury. We selected rapamycin based on accumulating evidence that CSD disrupts mTOR signaling, and aberrant mTOR activation contributes to neuronal apoptosis, oxidative stress, and synaptic dysfunction in sleep-deprived brains [39, 40]. As a well-characterized mTOR inhibitor, rapamycin provides a controllable and reproducible in vitro model to mimic downstream effects of mTOR dysregulation under CSD conditions. Moreover, previous studies have shown that rapamycin induces autophagy-associated apoptosis and oxidative stress in neuronal cells, partially recapitulating the cellular injury mechanisms relevant to CSD [41, 42]. Our study reveals that Rap-induced inhibition of the mTOR pathway and METTL3 knockdown enhance cleaved caspase-3 and Bax levels, thereby promoting apoptosis. Moreover, METTL3 knockdown further exacerbates these effects, suggesting that METTL3 may regulate neuronal survival and apoptosis in the hippocampus, potentially contributing to the pathology of cognitive deficits associated with CSD.

To further investigate the underlying mechanisms of METTL3 in hippocampal neurons, we performed sequencing analysis on HT-22 cells treated with METTL3 knockdown. Using WGCNA, correlation analysis, and differential expression analysis, we identified nine candidate genes linked to METTL3 from key module genes and DEGs. Enrichment and expression analyses of these candidate genes led us to select CDKN1A, a downstream gene with higher expression levels, for further investigation. Previous studies have shown that METTL3 deficiency promotes the migration, proliferation, and drug resistance of breast cancer cells by modulating the CDKN1A/AKT pathway, epithelial-mesenchymal transition (EMT), and the m⁶A-BAX/caspase-9/-3/-8 signaling pathways [43, 44]. Moreover, Yao FY et al. have reported that in chronic myeloid leukemia cells, METTL3 modulates the expression of NEAT1 through m⁶A modification, and that upregulation of miR-766-5p inhibits NEAT1 function [45]. Interestingly, CDKN1A is a miR-766-5p target gene, and CDKN1A knockdown can reverse the impacts of NEAT1, suggesting that the METTL3/NEAT1/miR-766-5p/CDKN1A axis is important in chronic myeloid leukemia. These outcomes highlight the importance of METTL3 in regulating CDKN1A expression. Our study found that the knockdown of METTL3 in the HT-22 cells led to an increased expression of CDKN1A, a reduction in m⁶A modification, and enhanced mRNA stability. These results suggest that METTL3 regulates hippocampal neuronal cell function by modulating CDKN1A expression, likely through its influence on m⁶A methylation and mRNA stability. Although we have not yet experimentally mapped the precise m⁶A sites on CDKN1A, there is evidence suggesting that this transcript may carry functional m⁶A modifications. For instance, POP1 has been shown to bind the CDS region of CDKN1A mRNA and promote its degradation in an m⁶A-dependent manner at position 497, a site recognized by YTHDF2 [46]. In addition, CDKN1A has been reported as a potential downstream target of METTL3 and may undergo m⁶A-mediated regulation in certain biological contexts [43, 47]. These findings imply that METTL3-recognizable m⁶A motifs are likely present on CDKN1A. In future work, m⁶A mapping assays would be performed to validate specific m⁶A sites on CDKN1A and to clarify how METTL3-dependent methylation regulates its stability and function in hippocampal neurons.

Cyclin Dependent Kinase Inhibitor 1A (CDKN1A) is a vital cell cycle regulator, primarily functioning as an inhibitor of cyclin-dependent kinases [48]. It is essential for preserving cell cycle arrest in reaction to DNA damage, thereby promoting cell survival or initiating apoptosis when necessary [49]. Recent studies demonstrate that CDKN1A is essential in cognitive impairment and neuronal dysfunction, as its overexpression can lead to neuronal cell cycle re-entry, causing neurodegeneration [50]. In Kumon H et al., haloperidol exposure during pregnancy resulted in the upregulation of apoptotic genes, including CDKN1A and APAF1, in the hippocampus of offspring, highlighting the potential impact of antipsychotics on neuronal development [51]. Similarly, Luyckx E et al. have reported that the loss of neuroglobin (Ngb) expression in neural stem cells (NSCs) led to increased proliferation, as indicated by the downregulation of CDKN1A and upregulation of CDK6, both of which promote cell cycle progression [52]. These findings suggest that alterations in CDKN1A expression can influence key cellular pathways, affecting neuronal and stem cell behavior.

Hippocampal damage or degeneration has been closely linked to various cognitive impairments and neurodegenerative diseases [53, 54]. Several studies have suggested that the hippocampus possesses a degree of self-repair capacity, especially following mild damage [55]. For instance, Konakanchi S et al. have reported that CSD led to impairments in spatial memory, increased anxiety-like behaviors, and a reduction in dendritic branching of CA3 hippocampal neurons in rats [56]. However, after 21 days of sleep recovery, improvements in memory, anxiety, and oxidative stress were observed. Similarly, Xie G et al. have found that sleep deprivation resulted in cognitive decline and hippocampal neuronal damage, with low-dose caffeine improving cognitive function while high-dose caffeine exacerbated cognitive damage [57]. In the current investigation, we analyzed the levels of METTL3, CDKN1A, and apoptosis-related proteins in the hippocampus of CSD-treated and METTL3-overexpressing rats. Our outcomes demonstrated that METTL3 overexpression alleviated the reduction of METTL3 caused by CSD, suppressed CDKN1A, and decreased the apoptosis-related proteins expression. Furthermore, MWM tests and histological analysis of hippocampal CA3 neurons indicated that METTL3 overexpression could restore cognitive function and neuronal integrity in CSD rats. Additionally, in hippocampal neurons, silencing METTL3 was confirmed to induce apoptosis through the upregulation of CDKN1A. These outcomes imply that METTL3 provides protection in the hippocampus, potentially via modulation of CDKN1A expression and apoptotic pathways, highlighting its therapeutic potential in mitigating cognitive decline and neurodegeneration.

Several limitations of this study should be acknowledged. First, only male rats were included, which may limit the generalizability of our findings to females. Second, the sample size was relatively modest, potentially reducing statistical power and the robustness of certain analyses, such as WGCNA. Third, HT-22 cells are an immortalized hippocampal neuronal line and may not fully recapitulate the physiological properties of primary neurons, particularly regarding synaptic protein expression. Key functional results therefore require validation in primary neuronal cultures. Fourth, our in vitro experiments cannot entirely replicate the complex in vivo conditions of chronic sleep deprivation. Finally, while we demonstrate that METTL3 regulates CDKN1A stability via m⁶A, the precise m⁶A modification motif has not yet been identified, and the role of m⁶A readers such as YTHDF2 remains to be explored.

To address these limitations and guide future research, studies should validate our findings in female animals and clinical samples to assess generalizability and translational relevance. Key functional results should be confirmed in primary neuronal cultures, and the roles of m⁶A motifs and readers such as YTHDF2 should be investigated. Additionally, exploring targeted drug interventions and evaluating their efficacy in preclinical trials will provide practical strategies for mitigating CSD-induced cognitive impairment. By acknowledging these limitations and outlining these directions, we aim to provide a balanced interpretation of our findings and inform subsequent studies.

Conclusion

This study suggests that METTL3 critically regulates cognitive function and neuronal survival under CSD conditions. Downregulation of METTL3 in CSD-induced model appears to promote cognitive deficits and neuronal apoptosis by modulating key target CDKN1A via m⁶A-dependent mechanisms. Notably, restoring METTL3 expression is associated with alleviates hippocampal damage and improves cognitive performance, underscoring its therapeutic potential. These findings identify the METTL3-m⁶A-CDKN1A axis as a molecular driver of CSD-related pathophysiology.

Author contributions

Conception and design of the research: Fei Xing, Xiao-Shan Shi and Jian-Jun Yang, Acquisition of data: Fei Xing, Xiao-Shan Shi and Min Jia, Analysis and interpretation of data: Fei Xing, Han-Wen Gu, Pan-Miao Liu and Lei Lei, Statistical analysis: Xiao-Shan Shi, Xing-Ming Wang and Mu-Huo Ji, Drafting the manuscript: Fei Xing and Xiao-Shan Shi, Revision of manuscript for important intellectual content: Jian-Jun Yang.

Funding

This study was supported by grants from Medical Science and Technology Research Project of Henan Province (No. SBGJ202403023), the National Natural Science Foundation of China (No. 82001187), the National Natural Science Foundation of China (No. U23A20421), and the Scientific Research and Innovation Team of The First Affiliated Hospital of Zhengzhou University (ZYCXTD2023012).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Competing interests

The authors declare no competing interests.

Consent for publication

Not applicable.

Supplementary information

The online version contains supplementary material available at 10.1038/s41398-026-03855-4.