What this is
- This research investigates the effects of () on cognitive function and gene expression in a mouse model of obstructive sleep apnea syndrome (OSAS).
- Mice were exposed to or () to simulate OSAS conditions, and their cognitive abilities were assessed through behavioral tests.
- RNA sequencing and quantitative PCR were used to analyze gene expression changes associated with cognitive dysfunction.
Essence
- impairs memory and learning in mice, linked to gene expression changes affecting mitochondrial function and oxidative stress response.
Key takeaways
- specifically induces cognitive dysfunction, as shown by impaired memory in passive avoidance tests. Mice exposed to had a lower proportion reaching the cutoff compared to control and groups.
- RNA sequencing revealed 170 downregulated and 46 upregulated genes in the group compared to controls. This contrasts with the group, which showed 151 downregulated and 44 upregulated genes, indicating distinct molecular responses.
- Key pathways, particularly the KEAP1-NFE2L2 antioxidant pathway, were significantly suppressed in the group. This suppression likely contributes to oxidative stress and cognitive impairment observed in the study.
Caveats
- Behavioral and molecular analyses were limited, lacking additional cognitive tests and protein-level validation. This limits the robustness of the findings.
- The study did not explore neurobiological changes such as neuronal morphology, which are important for understanding the full impact of on cognitive function.
- Absence of intermediate timepoints restricts understanding of acute versus chronic effects of hypoxia exposure on cognitive outcomes.
Definitions
- intermittent hypoxia (IH): Cyclical exposure to low oxygen levels followed by reoxygenation, mimicking conditions in obstructive sleep apnea.
- sustained hypoxia (SH): Continuous exposure to low oxygen levels without reoxygenation, used as a comparison to intermittent hypoxia.
- KEAP1-NFE2L2 pathway: A critical antioxidant defense pathway that regulates cellular responses to oxidative stress.
AI simplified
1. Introduction
Obstructive sleep apnea syndrome (OSAS) is the most common sleep-related breathing disorder, with a prevalence of 13–33% in men and 6–19% in women [1], affecting approximately 1 billion people worldwide [2], with this prevalence continuing to rise due to increasing obesity rates and population aging [3]. The fundamental pathophysiological feature of OSAS is recurrent upper airway obstruction during sleep, leading to cycles of hypoxia followed by reoxygenation [4].
OSAS presents diverse clinical manifestations, including snoring, excessive daytime sleepiness, and choking sensations during sleep. OSAS significantly affects multiple organ systems and increases the risk of hypertension [5], type 2 diabetes [6], and cardiovascular diseases [7]. Among these complications, cognitive dysfunction has emerged as an important complication of OSAS [8]. Patients with OSAS experience impaired attention, memory deficits, learning difficulties, and executive dysfunction [9,10]. Moreover, OSAS has been associated with increased risk of mild cognitive impairment and dementia [11].
The hippocampus plays a central role in memory formation and consolidation. Neuroimaging studies have revealed hippocampal atrophy in OSAS patients, correlating with decreased memory and learning functions [12]. These structural and functional changes represent a significant clinical burden, as cognitive dysfunction substantially impacts patients’ daily functioning and quality of life. At the cellular level, OSAS impairs synaptic plasticity in the hippocampal CA1 region and reduces neuronal numbers in both CA1 and CA3 regions [13,14]. However, the precise molecular mechanisms underlying OSAS-induced cognitive dysfunction remain poorly understood.
Intermittent hypoxia (IH) is a core pathophysiological condition in OSAS. In mouse IH models, cycling oxygen concentrations are used to reproduce hypoxia-reoxygenation cycles that mimic OSAS. Mice exposed to IH have shown impairments in spatial cognition, working memory, and learning in behavioral tests such as the Morris water maze test [15], Y-maze test [16], and passive avoidance test [17]. These cognitive dysfunctions involve not only decreased oxygen utilization due to hypoxia but also oxidative stress resulting from increased production of reactive oxygen species (ROS) during reoxygenation [18,19,20].
While IH-induced cognitive dysfunction has been demonstrated, studies that distinguish and evaluate the effects of hypoxia and reoxygenation in IH are limited, and the precise roles of specific genes and molecular pathways that cause cognitive dysfunction in the hippocampus remain unclear. Previous studies have established that repeated hypoxia-reoxygenation cycles increase ROS production [18], with sustained hypoxia (SH) typically leading to adaptive responses while intermittent hypoxia generates repeated oxidative stress that may overwhelm cellular antioxidant defenses [21,22,23,24].
Our working hypothesis was that cyclical reoxygenation in IH, rather than hypoxia per se, drives cognitive dysfunction through enhanced oxidative stress generation and suppression of antioxidant pathways. We anticipated that IH would result in greater pathway suppression, more pronounced gene expression changes, and more severe cognitive impairment compared to SH alone. To test this hypothesis, we employed both IH and SH models to investigate the differential effects of reoxygenation on cognitive function and hippocampal gene expression, aiming to elucidate the molecular mechanisms underlying OSAS-related cognitive dysfunction.
2. Results
2.1. Y-Maze Test and Passive Avoidance Test
Figure 1A shows the alternation rate in the Y-maze test. One-way analysis of variance showed no statistically significant effect of exposure conditions (F(2,46) = 1.460, p = 0.243). Figure 1B shows the proportion of mice that reached 300 s without entering the dark side of the chamber on day 2 of the passive avoidance test [control: 84.4% (95% CI: 68.2–93.1%), IH: 36.4% (95% CI: 15.2–64.6%), SH: 88.2% (95% CI: 65.7–96.7%)]. A chi-square test among the three groups revealed statistically significant differences between groups (χ2(2) = 10.719, p = 0.005). Pairwise comparisons using chi-square tests with Bonferroni correction showed that the IH group had a significantly lower proportion of mice reaching 300 s compared to both the control group (χ2(1) = 8.760, corrected p = 0.009) and the SH group (χ2(1) = 8.429, corrected p = 0.011). No significant difference was observed between the control and SH groups (χ2(1) = 0.139, p = 0.710).
2.2. Comparisons of Differentially Expressed Genes (DEGs)
To investigate the effects of IH and SH on gene expression in the mouse hippocampus, RNA sequencing (RNA-seq) was performed. Volcano plots of DEGs are shown in Figure 2A–C. Key genes of interest are labeled in the volcano plots, including Lars2, Hmcn1, Vstm2l, and Rps21, which were subsequently validated by RT-qPCR. Compared to the control group, the IH group showed 170 downregulated and 46 upregulated genes, while the SH group showed 151 downregulated and 44 upregulated genes. Direct comparison between IH and SH groups revealed 37 downregulated and 19 upregulated genes in the IH group relative to the SH group. The top 50 DEGs in each comparison, ranked by absolute log2 fold change (FC) magnitude, are shown in Table 1.
Figure 2D shows a Venn diagram of the DEGs. Venn diagram analysis revealed that 73 genes were shared between IH/control and SH/control comparisons, including Pknox1 and Klf2. Sixteen genes were shared between IH/control and IH/SH comparisons, including Adrb1. Twelve genes were shared between IH/SH and SH/control comparisons, including Dbi. Two genes (Cebpb and Gng13) were common to all three comparisons.
2.3. Functional Enrichment Analysis
Gene Ontology (GO) enrichment analysis using Metascape revealed significantly altered biological processes associated with DEG patterns (Figure 3). For DEGs in the IH vs. control comparison, enriched processes included neurological functions such as learning or memory (GO: 0007611, −log10 (p) = 3.86) and brain development (GO: 0007420, −log10 (p) = 2.96), as well as response to reactive oxygen species (GO: 0000302, −log10 (p) = 3.17), blood vessel development (GO: 0001568, −log10 (p) = 3.34), and IL-17 signaling pathway (mmu04657, −log10 (p) = 2.75) (Figure 3A). In contrast, DEGs in the SH vs. control comparison showed enrichment only in response to hypoxia (GO: 0001666, −log10 (p) = 2.22) with no effects on learning or memory (Figure 3B). The IH vs. SH comparison also revealed no enrichment in learning or memory (Figure 3C).
QIAGEN Ingenuity Pathway Analysis (IPA) (QIAGEN, Hilden, Germany) identified major molecular pathways altered in each comparison. Results are displayed as positive or negative Z-scores indicating activation and inhibition, respectively. Figure 4A,B show the top 20 activated and inhibited pathways in the IH group compared to controls, respectively. Activated pathways included mitochondrial dysfunction (−log10 (p) = 19.6, Z-score = 4.838), while inhibited pathways included the KEAP1-NFE2L2 antioxidant pathway (−log10 (p) = 5.74, Z-score = −5.303). For the IH vs. SH comparison, the top 20 altered pathways are shown in Figure 4C, D, with the KEAP1-NFE2L2 pathway inhibited (−log10 (p) = 4.33, Z-score = −3.441).
2.4. RT-qPCR
mRNA expression changes obtained by RT-qPCR are shown in Figure 5 and Table 2. Of the 23 genes analyzed, 19 genes showed no statistically significant effects of exposure conditions, including Adrb1, Foxo6, Trem2, Klf2, Sod3, Cebpb, Dbi, Manf, Trpc6, Hspa5, Nov, Basp1, H1fx, Ism1, Phc3, Pknox1, Rasl11a, Rps27, and Sdf2l1.
Four genes showed significant expression changes: Lars2 (F(2,22) = 4.376, p = 0.025), Rps21 (F(2,22) = 3.870, p = 0.036), Hmcn1 (F(2,22) = 9.196, p = 0.001), and Vstm2l (F(2,22) = 5.509, p = 0.012). Post-hoc analysis using Tukey’s HSD test revealed the following patterns: Lars2 expression was significantly lower in the IH group compared to the SH group (p = 0.036) with a trend toward decrease vs. control (p = 0.056). Rps21 expression was significantly lower in the SH group compared to control (p = 0.033). Hmcn1 expression was significantly lower in the IH group compared to both control (p = 0.008) and SH groups (p = 0.002). Vstm2l expression was significantly lower in the IH group compared to control (p = 0.012) with a trend toward decrease vs. SH (p = 0.055).
3. Discussion
The present study demonstrates that IH induces learning or memory impairments. RNA-seq and RT-qPCR analyses indicated the involvement of the KEAP1-NFE2L2 pathway and identified novel genes potentially associated with OSAS.
Behavioral tests demonstrated that IH specifically induces cognitive dysfunction. The passive avoidance test revealed significant learning or memory impairments in the IH group, with a decreased proportion of mice exceeding the cutoff value compared to both control and SH groups, while no difference was observed between SH and control groups. These findings are consistent with previous research using similar IH conditions (minimum FiO2 5%/4 min cycles/8 h daily/3 weeks) [17] and clearly indicate that intermittent, rather than sustained, hypoxia specifically impairs cognitive function. The differential sensitivity between these tests suggests that fear-motivated learning is more vulnerable to IH-induced oxidative stress than spatial working memory. This is likely because fear memory consolidation requires hippocampal neural circuit function [25], and the hippocampus is vulnerable to oxidative stress from repeated reoxygenation cycles [14]. Although the Y-maze test showed no significant differences between groups under our 4-week experimental conditions, this aligns with previous studies [17], and longer exposure durations may be required for spatial cognitive impairment [16].
RNA-seq analysis identified DEGs among groups. GO analysis of RNA-seq data revealed alterations in “learning or memory” and “response to reactive oxygen species” as key categories specifically altered in the IH group. Based on these results, RT-qPCR analysis revealed significant downregulation in four genes: Lars2, Vstm2l, Hmcn1, and Rps21. Lars2 encodes mitochondrial leucyl-tRNA synthetase essential for mitochondrial protein synthesis and functional maintenance [26,27,28,29]. This gene has been linked to Alzheimer’s disease, with knockdown studies demonstrating direct effects on neuronal function, including shortened axon length, reduced dendritic branching, increased mitochondrial superoxide levels, and neuronal cell death, and knockout mice exhibit cognitive dysfunction, increased p-tau, and hippocampal atrophy [30]. These neuromorphological alterations suggest that Lars2 downregulation may contribute to the cognitive deficits observed in our study. Additionally, decreased Lars2 expression impairs mitochondrial stress responses, leading to elevated ROS levels [31]. Vstm2l is localized to mitochondria and involved in maintaining mitochondrial homeostasis [32]. While specific roles of Vstm2l in synaptic plasticity remain to be fully elucidated, mitochondrial function is essential for neuronal energy metabolism. The previous study showed that repeated cycles of hypoxia-reoxygenation increased ROS production [18]. In addition, the increase in mitochondrial ROS is associated with neuronal cell death and cognitive dysfunction [19,20]. Therefore, the downregulation of mitochondrial genes Lars2 and Vstm2l suggests potential mechanisms for cognitive dysfunction that may involve impaired mitochondrial function and elevated ROS levels. Hmcn1, encoding an extracellular protein involved in epithelial cell junction organization [33], has been implicated in Alzheimer’s disease pathways [34]. In contrast, Rps21, encoding ribosomal protein S21 [35], showed downregulation specifically in the SH group. However, the preserved cognitive function in SH mice suggests that Rps21 downregulation alone may not be sufficient to cause behavioral deficits.
A key finding of this study was the significant suppression of the KEAP1-NFE2L2 pathway identified through QIAGEN IPA, with suppression observed in both IH and SH groups compared to the control group and notably stronger suppression in the IH group. The KEAP1-NFE2L2 pathway functions as a major regulator of intracellular antioxidant defense systems, playing cytoprotective roles including suppression of inflammatory signals [21], regulation of mitochondrial function [22], prevention of cell death [23], and neuroprotection [24]. Chronic IH exposure has been shown to decrease Nrf2 expression in multiple organs, including the hippocampus [14]. In the present study, although Nrf2 expression showed no significant differences between groups, pathway analysis revealed alterations in the KEAP1-NFE2L2 pathway. IH-induced suppression of the KEAP1-NFE2L2 pathway suggests potential attenuation of antioxidant defense and neuroprotective mechanisms, which may be associated with cognitive dysfunction. Regarding the KEAP1-NFE2L2 pathway, SH typically activates this pathway acutely but may lead to adaptive responses under chronic conditions. In contrast, IH involves repeated reoxygenation cycles that generate additional oxidative stress, resulting in greater pathway suppression than SH alone.
The coordinated downregulation of mitochondrial genes (Lars2, Vstm2l) alongside KEAP1-NFE2L2 pathway suppression suggests potential molecular pathways whereby IH exposure may affect cellular antioxidant defenses and mitochondrial function, possibly contributing to cognitive impairment. Figure 6 summarizes the proposed molecular mechanisms linking IH-induced ROS production to cognitive dysfunction through KEAP1-NFE2L2 pathway suppression and gene expression changes.
This study has several limitations. First, our behavioral and molecular analyses were limited in scope—additional cognitive tests, protein-level validation, and ROS measurements would strengthen our findings. Second, we did not examine neurobiological changes such as neuronal morphology or synaptic markers. Third, our study lacks intermediate timepoints to distinguish acute versus chronic effects of hypoxia exposure. Additionally, we did not perform correlation analysis between individual gene expression and pathway suppression scores. Future studies should include comprehensive validation approaches and multiple assessment timepoints to provide a more complete understanding of IH-induced cognitive dysfunction mechanisms.
4. Materials and Methods
4.1. Animals
Seventy-one male C57BL/6J mice (weight: 22.5 ± 2.9 g; age: 8 weeks; Japan SLC, Hamamatsu, Japan) were used for behavioral and molecular analyses. Mice were housed in groups of 5–6 per cage (30 cm length × 20 cm width × 12 cm height) under controlled environmental conditions at a temperature of 24 ± 1 °C and a 12:12 h light-dark cycle (lights on at 08:00). Mice had ad libitum access to food and water. All experimental protocols were approved by the Animal Research Committee of Showa Medical University (Tokyo, Japan) (approval code: 124040, approval date: 1 April 2024).
4.2. Experimental Protocol
Mice were allocated to the control group (n = 38), IH group (n = 16), and SH group (n = 17). The IH group mice were exposed to intermittent hypoxic loading with alternating oxygen concentrations of 10% and 21% in 2 min cycles for 8 h daily (12:00–20:00). The SH group mice were exposed to continuous hypoxic loading at 10% oxygen concentration for 8 h daily (12:00–20:00). The control group mice were housed without hypoxic exposure. Mice were maintained under these conditions for 28 days, followed by Y-maze testing on day 29, passive avoidance testing (day 1) on day 29, and passive avoidance testing (day 2) on day 30. After behavioral experiments, mice were euthanized under isoflurane anesthesia and decapitated for hippocampal collection. Hippocampi were immersed in RNA stabilization solution (RNAprotect Tissue Reagent, QIAGEN) to prevent RNA degradation, kept at 4 °C overnight, and then stored at −80 °C. Hippocampal samples were used for RNA-seq and cDNA synthesis.
4.3. Intermittent Hypoxia Exposure, Sustained Hypoxia Exposure
The 8 h hypoxia exposure period (12:00–20:00) during the light phase was selected based on established protocols in intermittent hypoxia research [36,37] and specifically following 8 h exposure protocols for cognitive function studies [14,17], and to mimic the inactive period in mice, corresponding to the sleep period when OSA occurs in humans. This timing aligns with the nocturnal circadian rhythm of mice, where the light phase represents their natural rest period.
The IH parameters were designed to approximate the pathophysiological patterns observed in human OSA. Clinical polysomnographic studies show that OSA patients experience desaturation-reoxygenation cycles with non-sigmoidal patterns, featuring faster reoxygenation compared to desaturation phases [38]. Our 2 min cycle protocol (70 s hypoxia, 50 s normoxia) was chosen to reflect these clinically observed temporal dynamics, providing a reasonable experimental model for studying IH-related effects. Mice in the IH group were exposed to intermittent hypoxia for 8 h daily from 12:00 to 20:00 for 28 days using a custom-made gas control delivery system…”Mice in the IH group were exposed to intermittent hypoxia for 8 h daily from 12:00 to 20:00 for 28 days using a custom-made gas control delivery system (Shibata Scientific Technology Ltd., Tokyo, Japan) with oxygen and nitrogen alternation to create alternating hypoxic and normoxic conditions, controlled by a pressure controller (Gas Cylinder Auto Changer Model 8500, WAKEN BTECH Co., Ltd., Kyoto, Japan). One cycle was defined as 120 s, consisting of a hypoxic phase (10% oxygen, 70 s) and a normoxic phase (21% oxygen, 50 s). Oxygen concentrations were continuously monitored by an oxygen analyzer (XP-3380II, Shin-Cosmos Electric Co., Ltd., Osaka, Japan) and recorded using an A/D converter (PL2604 PowerLab 4/26, ADInstruments, Dunedin, New Zealand) connected to LabChart v7 (ADInstruments).
Mice in the SH group were housed using a custom-made acrylic box (56 cm width × 45 cm height × 41 cm depth) (Kyodo International Co., Ltd., Kawasaki, Japan) with their cages placed inside. The box featured five 0.8 cm air holes: two at the upper corners of the back panel and two at the lower corners of each side panel, with a fan attached at a height of 28 cm on the left side panel that operated continuously throughout the experiment. A gas control delivery system (ProOx P110, BioSpherix, Parish, NY, USA) and a pressure controller (Gas Cylinder Auto Changer Model 8500, WAKEN BTECH Co., Ltd.) were employed to maintain hypoxic conditions using nitrogen. The system continuously monitored oxygen concentration using a built-in oxygen analyzer and featured a programmable timer function (H5S, OMRON, Kyoto, Japan). Mice were exposed to sustained hypoxia at 10% oxygen concentration for 8 h daily (12:00–20:00) over 28 days. During the remaining period (20:00–12:00), atmospheric oxygen concentration (21%) was maintained using an air pump (AP-30P, Yasunaga Air Pump Co., Ltd., Tokyo, Japan).
Both gas delivery systems were validated before experiments and underwent regular calibration to ensure accurate gas circulation and maintain stable concentrations.
4.4. Y-Maze Test
Based on previous studies [16,17,39], mouse short-term spatial working memory was evaluated using the Y-maze test. Forty-nine mice (control group: n = 17, IH group: n = 16, SH group: n = 17) were analyzed, with one mouse from the IH group excluded due to deviation from the Y-maze test during the experiment. The maze consisted of three identical arms radiating from a central area at 120° angles to each other, with dimensions of 40 cm length × 12 cm height, 3 cm width at the floor expanding to 10 cm width at the ceiling. Arm entry was defined as all four paws completely entering an arm. Mice explored the maze for 8 min while the number of arm entries was recorded by an overhead video camera (MX Brio C1100PG, Logitech International S.A., Lausanne, Switzerland) and analyzed using behavioral analysis software (SMART V 3.0, Bio Research Center Co., Ltd., Nagoya, Japan). The percentage of spontaneous alternation was calculated as: %Alternations = (number of triads/(N − 2)) × 100, where N = total number of entries.
4.5. Passive Avoidance Test
Sample size calculations were based on standard guidelines for χ2 analysis. For detecting a large effect size (Cohen’s h ≥ 0.7) with α = 0.05 and power = 0.80, the minimum required sample size is approximately n = 16 per group. Based on previous studies [17,40], mouse memory and learning ability were evaluated using a two-compartment step-through passive avoidance apparatus (MPB-M030, Melquest Co., Toyama, Japan). Seventy-one mice (control group: n = 38, IH group: n = 16, SH group: n = 17) were initially used. The apparatus consisted of a bright compartment (10.0 × 18.0 × 14.5 cm) and a dark compartment (18.0 × 18.0 × 14.5 cm) separated by a wall with a guillotine door. The bright compartment was illuminated at 145 lx.
During the training phase, mice were placed in the bright compartment for 20 s before the guillotine door was opened to allow entry into the dark compartment. When mice entered the dark compartment, the guillotine door was closed, and an electric shock (0.3 mA) was delivered for 3 s. The 0.3 mA shock intensity was selected based on manufacturer recommendations (Melquest Co.) and is consistent with established protocols using 0.3 mA for 3 s in passive avoidance testing [40], ensuring reliable learning responses while maintaining appropriate animal welfare standards. Testing was performed 24 h after training. Mice were placed in the bright compartment for 20 s before the guillotine door was opened. The latency to enter the dark compartment was recorded for up to 300 s. Mice that did not enter the dark compartment after 300 s were considered to have retained memory, and this proportion was used for analysis. Due to experimental protocol errors, 11 mice (6 from the control group and 5 from the IH group) that received multiple electrical stimulations were excluded from the analysis, resulting in final group sizes of n = 32 (control), n = 11 (IH), and n = 17 (SH).
4.6. RNA Sequencing
Total RNA was extracted from hippocampi using the RNeasy® Plus Universal Mini Kit (QIAGEN) following the manufacturer’s protocol. Total RNA concentration and purity were determined based on the ratio of absorbance at 260 nm and 280 nm using a NanoDrop One Spectrophotometer (Thermo Fisher Scientific, Waltham, MA, USA). RNA samples with A260/280 ratios ≥ 1.8 and A260/230 ratios ≥ 1.8 were considered acceptable for further analysis. For RNA-seq analysis, 5 samples each from the IH and SH groups and 4 samples from the control group were submitted to Rhelixa Co., Ltd. (Tokyo, Japan). Control samples were prepared from mice exposed to atmospheric conditions under identical experimental conditions (same mouse strain, age, and housing duration) in a preliminary experiment. RNA integrity was assessed by Rhelixa Co., Ltd., and only samples with RNA Integrity Number ≥ 8.0 were used for sequencing. Library preparation was performed using the NEBNext Ultra II Directional RNA Library Prep Kit with NEBNext Poly(A) mRNA Magnetic Isolation Module for poly-A selection (New England Biolabs, Ipswich, MA, USA). Sequencing was conducted on an Illumina NovaSeq 6000 platform (Illumina, San Diego, CA, USA), generating paired-end 150 bp reads with an average sequencing depth of 26.7 million reads per sample.
In the primary analysis, adapter sequences and low-quality bases in paired-end reads were removed with fastp (version 0.23.4). Filtered paired-end reads were mapped to the mouse reference genome (GRCm39) by HISAT2 (version 2.2.1) and expression levels were quantified by StringTie (version 2.2.1). The read count values of the gene expression data (raw signal) were subjected to appropriate processing by Subio Platform (Subio Inc, Nagoya, Japan), including binary log transformation, global normalization by 80th percentile, low signal cutoff (read counts < 40), missing value completion (log2 32), correction by control group values (Processed signal), averaging and ratio calculation to the control group by exponential function (FC). Specifically, among the read gene expression data (40,015 genes), only genes for which the raw signal in each group was above the low signal cutoff value for protein-coding genes were extracted (13,733 genes) and further filtered to genes with a processed signal ≥ |0.25| (7756 genes).
For secondary analysis, functional analysis was performed using Metascape to identify enriched biological pathways and processes [41]. DEG lists were analyzed with thresholds of FC > 1.5 or < 0.67 and significance levels with p < 0.05. Furthermore, pathway analysis was conducted using QIAGEN IPA to examine molecular networks. Of the 7756 genes, 38 genes that were not mapped within the QIAGEN IPA were excluded, and of the remaining 7718 genes, those meeting p < 0.05 were analyzed. Data were considered significant with thresholds of −log10 (p) > 1.3 and |Z-score| > 2.0. The sequence data were deposited in the Gene Expression Omnibus database (https://www.ncbi.nlm.nih.gov/geo↗, accessed on 25 July 2025; accession no. GSE 299437).
4.7. RT-qPCR
For RT-qPCR analysis, total RNA samples from the control group included 9 samples extracted in this experiment, while the IH and SH groups used 8 samples from each group. cDNA was synthesized from total RNA using PrimeScript™ RT Master Mix (Takara Bio Inc., Kusatsu, Japan) according to the manufacturer’s protocol.
Target genes were selected based on several criteria. Based on GO analysis results, genes with read counts >100 that have been reported to be associated with cognitive function or neural processes within the “learning or memory” and “response to reactive oxygen species” categories were selected (Adrb1 [42], Foxo6 [43], Trem2 [44], Klf2 [45], Sod3 [46]). Additionally, three categories of genes were included: genes previously reported to be associated with cognitive function or neural processes but not identified in the GO analysis (Cebpb [47], Dbi [48], Lars2 [30], Manf [49], Trpc6 [50]), genes with high read counts (>1500) but no existing literature reports (Hspa5, Nov, Rps21), and genes with large FC (>2.0 or <0.5) (Basp1, H1fx, Hmcn1, Ism1, Phc3, Pknox1, Rasl11a, Rps27, Sdf2l1, Vstm2l).
Probe details are provided in Table 3. All probes contained 6-FAM at the 5′ end, ZEN quencher internally, and Iowa Black FQ quencher at the 3′ end and were synthesized through a PrimeTime Mini qPCR Assay (Integrated DNA Technologies, Inc., Coralville, IA, USA). Quantitative PCR was performed in a total volume of 10 μL with PrimeTime® Gene Expression Master Mix (Integrated DNA Technologies, Inc., Coralville, IA, USA) and QuantStudio™ 5 Real-Time PCR system (Thermo Fisher Scientific, Waltham, MA, USA) according to the manufacturer’s instructions. The PCR conditions were as follows: initial denaturation at 95 °C for 3 min, followed by 45 cycles of 95 °C for 15 s and 60 °C for 30 s.
The relative expression levels of each gene were determined by the ΔΔCt method using Gapdh as an internal control. Gapdh was selected as the reference gene based on its established stability in intermittent hypoxia studies. Previous validation studies using RT-qPCR have demonstrated Gapdh stability in chronic intermittent hypoxia models of sleep apnea in both brain tissue [51] and cardiac tissue [52], supporting its use as an appropriate internal control for this experimental paradigm.
4.8. Statistical Analysis
Statistical analyses were performed using JMP Pro ver.17.0.0 (JMP Statistical Discovery LLC., Cary, NC, USA). All data are presented as mean ± standard error. For Y-maze test alternation behavior rates and hippocampal mRNA levels, one-way analysis of variance (ANOVA) followed by Tukey’s HSD test was used for multiple group comparisons. For the passive avoidance test, the proportion of mice exceeding the cutoff value (300 s) was compared using the chi-square test. Statistical significance was set at p < 0.05.
5. Conclusions
In this study, we demonstrated that IH induces learning or memory impairments using a mouse model of OSAS. RNA-seq and RT-qPCR analyses revealed three genes (Lars2, Hmcn1, and Vstm2l) that showed specific downregulation in the IH group and confirmed the involvement of the KEAP1-NFE2L2 antioxidant pathway. These findings provide insights into the molecular basis of cognitive dysfunction in OSAS.