Identification of Oral Protein Biomarkers for Cognitive Impairment Susceptibility in Obstructive Sleep Apnea: A Cross-Species Proteomic Study

Abstract

Obstructive sleep apnea-hypopnea syndrome (OSAHS) is a common sleep disorder, but cognitive impairment occurs only in a subset of patients, indicating individual susceptibility. This study aimed to identify oral fluid-derived protein biomarkers potentially related to neurobehavioral vulnerability in OSAHS by integrating clinical and animal model data. Nineteen participants (13 OSAHS, 6 controls) provided gingival crevicular fluid; age, body mass index (BMI), and weight showed no significant differences, while OSAHS patients exhibited significantly higher Epworth Sleepiness Scale (ESS) scores, reflecting increased daytime sleepiness. Fourteen C57BL/6J mice were randomly assigned to chronic intermittent hypoxia or normoxia conditions for 12 weeks; behavioral performance was evaluated using open field and Y-maze tests, and oral fluid was collected for proteomic analysis. 4D-DIA profiling identified 225 human and 105 mouse differentially expressed proteins; enrichment analysis highlighted humoral immunity and complement pathways. Notably, FN1 and JCHAIN were consistently upregulated across species, with expression changes accompanied by behavioral alterations in CIH mice. This clinic-driven, cross-species experimental study revealed FN1 and JCHAIN as shared, upregulated proteins potentially linked to hypoxia-associated neurobehavioral vulnerability in OSAHS. Rather than broadly focusing on differential expression, the study highlights these two proteins as candidates for further mechanistic investigation and future biomarker validation in cognitively vulnerable OSAHS patients.

Introduction

Obstructive sleep apnea-hypopnea syndrome (OSAHS) is a common sleep-related breathing disorder characterized by recurrent upper airway collapse during sleep, resulting in intermittent hypoxia and sleep fragmentation. These physiological disruptions contribute to widespread systemic effects, including metabolic dysregulation, cardiovascular complications, and notably, cognitive dysfunction. Accumulating evidence has linked OSAHS to impairments in attention, memory, and executive function, as well as an increased risk of neurodegenerative diseases such as Alzheimer’s disease (AD). The underlying mechanisms are thought to involve hypoxia-induced neuroinflammation, oxidative stress, and impaired synaptic plasticity, particularly affecting the hippocampus and prefrontal cortex.

However, not all individuals with OSAHS exhibit cognitive deficits, highlighting significant phenotypic heterogeneity within this population. Large-scale cohort studies have proposed symptom-based OSAHS subtypes, including the excessive daytime sleepiness phenotype, which has been linked to increased cardiovascular and neurocognitive risks. These findings suggest that cognitive vulnerability may characterize a distinct OSAHS subtype with specific biological underpinnings.

Despite these insights, early identification of cognitively vulnerable patients remains a clinical challenge due to the lack of reliable, noninvasive biomarkers. With the rapid advancement of omics technologies, proteomics has emerged as a powerful approach to explore complex disease mechanisms. , Existing studies have demonstrated that OSAHS is associated with metabolic disturbances, and altered protein expression profiles. Notably, noninvasively collected biofluids such as urine have shown promise in biomarker discovery. In animal models of chronic intermittent hypoxia (CIH), urinary markers such as trimethylamine N-oxide (TMAO) and allantoin have been linked to oxidative stress and systemic metabolic disruption, reflecting hypoxia-induced pathophysiology. Separately, longitudinal human studies have identified urinary proteins, including ORM1, ORM2, and SERPINA3, as predictors of cognitive decline in the elderly, independent of OSAHS status. Together, these findings suggest that urinary proteomics can reveal molecular signatures of both systemic hypoxic stress and progressive neurodegeneration. Oral fluids, such as saliva and gingival crevicular fluid (GCF), offer similarly noninvasive sampling advantages and have shown promise in reflecting both local and systemic physiological changes in sleep-related disorders. However, the proteomic characteristics of oral proteins in OSAHS with cognitive impairment remain largely unexplored.

In this study, we hypothesize that intermittent hypoxia in OSAHS may induce systemic metabolic alterations that are reflected in the oral environment. Our primary aim is to identify protein signatures associated with OSAHS and its daytime functional burden, with daytime sleepiness described by the Epworth Sleepiness Scale, a subjective measure of sleepiness. , As a secondary objective, we seek to characterize the functional categories of these proteins to gain insight into the biological mechanisms underlying neurobehavioral vulnerability.

Experimental Procedures

The study workflow is illustrated in Figure .

1.

Study workflow overview OSA: obstructive sleep apnea; CIH: chronic intermittent hypoxia; DIA: data-independent acquisition. Created in BioRender. Yu, V. (2025) https://BioRender.com/u5wbc19.

Human Clinical Study

Study Population

The study was carried out in the Department of Orthodontics at Peking University Hospital of Stomatology, registered with the Chinese Clinical Trial Registry (ChiCTR2000038751), with ethical clearance granted by the Institutional Review Board of Peking University School of Stomatology (PKUSSIRB-201950155). All procedures were conducted in accordance with the Declaration of Helsinki, and written informed consent was obtained from all participants prior to enrollment.

The clinical study was designed as a case-control study, with a 2:1 ratio in the OSAHS (n = 13) and control (n = 6) groups. The inclusion criteria for the OSAHS group were: (1) male, aged 18 to 60 years; (2) no use of medications affecting protein metabolism within the past 3 months; (3) no severe psychiatric disorders such as diagnostic anxiety or depression; (4) ESS score greater than 10; (5) Apnea–hypopnea index (AHI) greater than 5, with no previous treatment for OSAHS; (6) ability to comply with study procedures and provide written informed consent; (7) no history of orthodontic or periodontal treatment within the past 6 months. A total of 50 individuals with suspected OSA who provided informed consent were initially enrolled. All participants underwent overnight audio/video polysomnography (PSG) at the Sleep Division of Peking University People’s Hospital to confirm their sleep status, and 7 individuals with an AHI < 5 were excluded, as they did not meet diagnostic criteria for OSA. The remaining participants were further screened with the ESS, and 15 individuals with ESS < 10 were excluded due to the absence of marked daytime sleepiness. In addition, 1 participant with a diagnosed anxiety disorder was excluded to minimize potential confounding. Ultimately, 13 male patients with OSAHS were included in the study.

The inclusion criteria for the control group were: (1) matched age and sex with the OSAHS group; (2) no sleep-related complaints (snoring, insomnia, poor sleep quality, etc.); (3) no evidence of OSAHS or other sleep disorders in PSG.

Exclusion criteria for both groups included: (1) coexisting sleep disorders identified by PSG, such as central hypersomnia, periodic limb movement disorder, restless legs syndrome, or narcolepsy; (2) current smokers or individuals who had quit smoking within the past 6 months; (3) presence of infectious or communicable diseases; (4) use of antibiotics, antifungals, antivirals, antiparasitic drugs, corticosteroids, immunosuppressants, or immunomodulators within the past 3 months, regardless of route of administration; (5) special dietary habits (e.g., vegetarianism, ketogenic diet, etc.).

Polysomnography

Overnight video and audio PSG, incorporating full electroencephalographic (EEG) monitoring with a standard multichannel montage, was performed using the Compumedics E-series system (Compumedics, Abbotsford, Victoria, Australia) in accordance with the guidelines of the American Academy of Sleep Medicine (AASM). The PSG recording protocol followed standard procedures previously described, and were manually scored by certified sleep technologists and independently reviewed by a researcher who was blinded to participant group allocation and study outcomes. Sleep stages, arousals, and respiratory events were scored based on the AASM criteria. The AHI was calculated as the total number of apneas and hypopneas per hour of sleep. Diagnosis of OSAHS was established based on an AHI greater than 5 events per hour.

Gingival Crevicular Fluid Samples Collection

Gingival crevicular fluid (GCF) samples were collected in the morning following PSG examinations between 6:30 and 8:30 AM. Participants were instructed to refrain from eating, drinking, chewing gum, or performing any oral hygiene procedures for at least 2 h prior to collection. Before sampling, subjects rinsed their mouths with clean water and rested quietly for 10 min. Gingival crevicular fluid was collected by experienced clinicians using absorbent paper strips. For each sample, strips were gently inserted into six designated sites around a single tooth, specifically the mesial, middle, and distal regions of both the buccal and lingual surfaces. Prior to insertion, the target area was air-dried and supragingival plaque was carefully removed. Strips were inserted into the gingival sulcus until slight resistance was encountered, and left in place for 30 s to allow adequate absorption. Care was taken to avoid contamination with blood. Only the visibly moistened portion of each strip was retained. The visibly moistened portions were retained and immediately wrapped in aluminum foil, snap-frozen in liquid nitrogen, and stored at −80 °C until further analysis. All samples were subsequently processed and analyzed in a single batch to minimize potential batch effects.

Murine Experimental Study

Animal Model and Experimental Design

To investigate the mechanisms of OSAHS in the absence of an ideal clinical model, a chronic intermittent hypoxia (CIH) mouse model was established in this study. Behavioral assessments were used to evaluate cognitive function, and oral proteomic analysis was performed to identify proteins that overlapped with those found in clinical samples, providing a basis for further mechanistic exploration.

Eight-week-old male C57BL/6J mice were purchased from the Tongxiang Branch, Zhejiang Vital River Laboratory Animal Co. Ltd., China. All mice were acclimated for 1 week under standard conditions, with a 12-h light/dark cycle (lights on from 7:00 AM to 7:00 PM). Age- and background-matched control mice were maintained under identical housing conditions but were exposed to ambient air without hypoxia. All animal procedures were conducted in accordance with institutional guidelines and were approved by the Laboratory Animal Welfare and Ethics Committee of Peking University Health Science Center (Approval No. DLASBD0276).

After acclimatization, mice in the experimental group were subjected to chronic intermittent hypoxia (CIH) using an Oxycycler A84XOV system (Biospherix, Redfield, NY). The CIH protocol consisted of 5 min cycles: oxygen levels were decreased from 21% to 5% ± 1% over 150 s, maintained at 5% for 30 s, and then gradually returned to 21% ± 1% over 120 s. This cycle was repeated 12 times per hour, for 8 h daily (from 9:00 AM to 5:00 PM).

After 12 weeks of exposure, behavioral tests were conducted, and mice were euthanized. Body weights were recorded prior to euthanasia.

Behavior Test

To comprehensively evaluate the impact of CIH on locomotor and cognitive function, two behavioral tests were performed: the Open Field Test (OFT) and the Y-maze test. The OFT assessed spontaneous locomotor activity and exploratory behavior. The Y-maze test was employed to evaluate working memory and spatial memory, serving as a classic paradigm for assessing attention and cognitive ability. All behavioral tests were conducted once, after the completion of the modeling period and prior to euthanasia.

Open Field Test

The open field test was conducted in a quiet, dimly lit environment. Mice were acclimated to the behavioral testing room for 1 h prior to the experiment. Each mouse was gently placed in the center of a square open field apparatus (42 × 42 cm) and allowed to explore freely for 15 min. The entire session was video recorded to monitor locomotor activity. After each trial, the apparatus was thoroughly cleaned with 75% ethanol to eliminate residual odors. The central zone was defined as a square area (21 × 21 cm) centered in the middle of the arena, while the peripheral zone referred to the 21 cm-wide area along the walls. Measured parameters included time spent in the center and periphery, as well as average locomotion speed.

Y-Maze Spontaneous Alternation Test

After a 1-h acclimation period in the behavioral testing room, mice were placed in the center of a Y-maze with three identical arms labeled A, B, and C. Each mouse was allowed to explore freely for 6 min, and the sequence of arm entries was recorded. An arm entry was counted when all four limbs of the mouse entered an arm. Spontaneous alternation was defined as consecutive entries into three different arms (e.g., A–B–C). The percentage of spontaneous alternation was calculated as

$$\mathrm{a}\mathrm{lternation}\%=\frac{\mathrm{number}\mathrm{of}\mathrm{correct}\mathrm{alternations}}{\mathrm{total}\mathrm{arm}\mathrm{entries}-2}\times 100\%$$

Murine Oral Fluid Collection

Oral fluid collection was performed from afternoon to evening within a fixed time window that was kept consistent across animals. In mice, the oral cavity was rinsed twice using 500 μL of phosphate-buffered saline (PBS). Subsequently, the mandible, containing two incisors and six molars, was carefully extracted and immersed in 1.5 mL of PBS containing the previously collected oral rinse. The sample was then subjected to ultrasonic treatment at a power output of 60 W for 20 s to dislodge and disperse adherent microorganisms. Given the anatomical and procedural context, the resulting fluid likely contains a mixture of saliva, gingival crevicular components, and tissue exudates. For comparative purposes, this sample is hereafter referred to as GCF-like oral fluid. The resulting suspension was stored at −80 °C until further analysis. All samples were subsequently processed and analyzed in a single batch to minimize potential batch effects. Mice were euthanized via intraperitoneal injection of anesthetic overdose.

4D-DIA-Based LC-MS/MS Proteomics

Proteins were extracted from each sample using an appropriate volume of SDT lysis buffer (4% SDS, 100 mM Tris-HCl, pH 7.6), followed by quantification using the BCA assay. For each sample, 20 μg of total protein was mixed with 5× SDS loading buffer, boiled for 5 min, and subjected to SDS-PAGE (4–20% gradient precast gel, 180 V constant voltage, 45 min). The gels were stained with Coomassie Brilliant Blue R-250. Equal amounts of protein from all samples were pooled to generate a composite sample used for spectral library construction.

All samples, including the pooled composite, were digested with trypsin using the filter-aided sample preparation (FASP) protocol. The digested peptides from the pooled sample were fractionated into 10 fractions using the Thermo High pH Reversed-Phase Peptide Fractionation Kit. All fractions and individual sample digests were desalted using C18 cartridges, lyophilized, and reconstituted in 40 μL of 0.1% formic acid. Peptide concentration was determined by OD280 measurement. Indexed Retention Time (iRT) standard peptides were spiked into both pooled and individual peptide samples prior to LC-MS analysis.

For DDA (data-dependent acquisition) analysis, peptides were separated using a nanoflow Evosep One system (Evosep, Denmark) and analyzed on a timsTOF mass spectrometer (Bruker) operating in positive ion mode. MS parameters were as follows: spray voltage 1500 V, source temperature 180 °C, dry gas flow 3 L/min. PASEF cycles: 8; TIMS accumulation time: 100 ms; ion mobility range: 0.75–1.35 Vs/cm²; MS and MS/MS range: 100–1700 m/z; dynamic exclusion: 24 s.

For DIA (data-independent acquisition) analysis, the same LC-MS setup was used. MS/MS data were acquired in DIA mode with four TIMS scan windows, each with an accumulation time of 100 ms. In PASEF mode, collision energy was linearly ramped based on ion mobility 1/K₀ (from 20 to 59 eV across 1/K₀ = 0.85–1.30 Vs/cm²).

DDA raw files were processed using Spectronaut (version 14.4.200727.47784) to build a spectral library. A species-specific protein database was used, with the following search parameters: enzyme specificity set to trypsin with a maximum of one missed cleavage; fixed modification: carbamidomethylation (C); variable modifications: methionine oxidation and N-terminal acetylation. Protein identifications were filtered with a false discovery rate (FDR) threshold of <1%.

DIA data were analyzed using the same version of Spectronaut. The spectral library and database used were identical to those for DDA. Analysis parameters included dynamic iRT for retention time prediction, MS2 interference correction enabled, and cross-run normalization enabled. Peptide and protein identifications were filtered using a Q value cutoff of 0.01, corresponding to FDR < 1%. The mass spectrometry proteomics data generated in this study have been deposited to the ProteomeXchange Consortium via the iProX partner repository with the data set identifier PXD068807.

Bioinformatics Analysis

Significantly differentially expressed proteins were identified based on a fold change (FC) greater than 1.5 or less than 0.67, combined with a p-value below 0.05 (Welch’s t test or equivalent). To examine expression patterns within and between groups, and to evaluate whether these differences reflected meaningful biological effects, hierarchical clustering analysis was performed on the identified proteins. The clustering results were visualized using heatmaps to reveal sample grouping and expression trends. Subsequent functional annotation was conducted via Gene Ontology (GO) analysis using Blast2GO, focusing on level 2 categories across biological processes (BP), molecular functions (MF), and cellular components (CC). KEGG pathway enrichment was performed using the Kyoto Encyclopedia of Genes and Genomes and statistically significant enrichment was determined using Fisher’s exact test (p < 0.05).

Statistical Analysis

All statistical analyses were conducted using R version 4.3.0 (The R Foundation for Statistical Computing, Vienna, Austria). During the preprocessing of proteomic data, zero values were replaced with one-tenth of the smallest positive value, followed by logarithmic transformation and missing value control to meet the assumption of normal distribution. Intergroup comparisons were performed using Welch’s t test, which does not assume equal variances. Differential expression was defined by a p-value <0.05 and |log 2FC| > log 2(1.5). For clinical correlation analysis, normality testing was conducted, revealing that AHI, neck circumference, and lowest oxygen saturation did not follow a normal distribution; thus, Spearman’s correlation analysis was employed for these parameters. In contrast, Pearson’s correlation analysis was used to assess associations between protein expression and ESS, age, height, weight, and BMI. Statistical significance was set at an adjusted p-value <0.05 for enrichment analysis, and p < 0.05 was considered statistically significant throughout all tests.

Results

Human study

Study Population

This cross-sectional study enrolled 19 participants, comprising 13 individuals diagnosed with OSAHS and 6 healthy, nonsnoring controls. There were no statistically significant differences between the two groups in terms of age (controls: 38.0  ±  5.6 years; OSAHS: 34.3  ±  19.6 years), body mass index (BMI) (controls: 24.89; OSAHS: 28.70), height (controls: 1.74 m; OSAHS: 1.74 m), or weight (controls: 75.83 kg; OSAHS: 87.51 kg). The median AHI in the OSAHS group was 34.5 events per hour, with an interquartile range of 13.6 to 52.3, confirming the presence of clinically significant obstructive sleep apnea.

The OSAHS group had higher ESS scores (15.31 ± 4.15) than the control group (10.17 ± 4.54), and the difference was statistically significant (p = 0.028), which reflects an increase in perceived daytime sleepiness among individuals with OSAHS. The lowest oxygen saturation (SpO₂) recorded during PSG was also lower in the OSAHS group (mean: 79.2%; range: 48–91%) compared to the control group (mean: 92.6%; range: 91–95%), suggesting that patients with OSAHS experienced more frequent or severe episodes of nocturnal hypoxemia.

Identification and Expression Profile of Gingival Crevicular Fluid Proteins in OSAHS Patients

Using 4D-DIA technology, a total of 761 protein groups and 3650 peptides were identified in the GCF samples from OSAHS patients, with a FDR below 1% at the peptide level. Among the shared proteins, 225 were found to be significantly differentially expressed, including 122 upregulated and 103 downregulated proteins. The abnormal expression profile in OSAHS was visualized through volcano plots and heatmaps (Figure A,B).

2.

Identification and characterization of gingival crevicular fluid in OSAHS Patients by 4D-DIA proteomics analysis (A) and (B) Differentially expressed proteins between OSAHS and control groups illustrated by volcano plot (A) and heatmap (B); (C) Correlation analysis between differentially expressed proteins and clinical variables; (D) to (G) Dysregulated protein enrichment analysis revealed the potential molecular function (D), cellular component (E), biological process (F), and KEGG pathways (G) enriched in the OSAHS group compared to the control group. OSAHS: obstructive sleep apnea-hypopnea syndrome; *p < 0.05;**p < 0.01; ***p < 0.001.

Among the 225 significantly differentially expressed proteins, most were associated with at least one clinical variable. Several proteins such as PZP, APOA1, SERPINA7, and AFM (upregulated), and S100A8, S100A12, RANBP1, and HNRNPU (downregulated), were linked to three or more clinical features, suggesting that these proteins may serve as promising biomarker candidates related to OSAHS (Figure C).

Functional Characterization of GCF Proteins in OSAHS Patients

To investigate the potential biological roles of the DEPs, GO annotation and KEGG pathway enrichment analyses were performed.

GO molecular function analysis revealed significant enrichment in peptidase regulator activity and peptidase inhibitor activity (adjusted p < 0.001), implicating disruptions in proteolytic regulation (Figure D). Furthermore, complement binding and serine-type endopeptidase inhibitor activity were also enriched, reinforcing the potential role of immune modulation and inflammatory protein control in OSAHS. In terms of cellular components, the DEPs were primarily localized in blood microparticles (adjusted p < 0.001) and secretory granule lumen (adjusted p < 0.001), indicating secretion-related and extracellular vesicle involvement. Enrichment in immunoglobulin complex and cytoplasmic vesicle lumen further supports their role in immune responses (Figure E). Biological process analysis showed enrichment in complement activation (adjusted p < 0.001) and humoral immune response (adjusted p < 0.001) (Figure F). The involvement of immune-related biological processes in biological process analysis further supports the findings from molecular function and cellular components analyses, implicating immune regulation as a central mechanism. KEGG pathway analysis indicated that the DEPs were significantly enriched in the complement and coagulation cascades pathway (adjusted p < 0.001) (Figure G).

These results suggest that dysregulated immune-related pathways, particularly complement activation and coagulation cascades, may contribute to the underlying pathophysiology of OSAHS.

CIH Mouse Model and Proteomic Analysis

A total of 14 male C57BL/6J mice were included and divided into a CIH group (n = 7) or a normoxia control group (n = 7). The two groups were matched in age, sex, and housing conditions, and only differed in oxygen exposure. After 12 weeks of CIH, the body weight of mice in the CIH group was significantly lower than in the control group (p < 0.05).

Using 4D-DIA proteomic technology, a total of 4508 protein groups and 26,073 peptides were identified. Of these, 4051 protein groups were shared between CIH and control groups, with 257 proteins detected only in the CIH group and 104 only in the control group. Principal component analysis (PCA) revealed a clear separation between the two groups, indicating a significant effect of hypoxia on oral protein expression (Figure A).

3.

Identification and characterization of oral fluid in CIH mice by 4D-DIA proteomics analysis (A) The principal component analysis (PCA) of the CIH and the normoxia groups; (B) and (C) Differentially expressed proteins between the CIH and the normoxia groups illustrated by heatmap (B) and volcano plot (C); (D) to (G) Dysregulated protein enrichment analysis revealed the potential molecular function (D), cellular component (E), biological process (F), and KEGG pathways (G) enriched in the CIH group compared to the normoxia group. CIH: chronic intermittent hypoxia.

A total of 105 significantly differentially expressed proteins (adjusted p < 0.05) were identified, including 57 upregulated and 48 downregulated proteins. The abnormal expression profile induced by CIH was visualized using heatmaps and volcano plots (Figure B,C).

Functional Characteristics of Differentially Expressed Oral Proteins in CIH Mice

In terms of molecular function, DEPs were significantly enriched in ankyrin binding (adjusted p = 0.0087) and structural constituent of cytoskeleton (adjusted p = 0.0087) (Figure D). Cellular component analysis showed that DEPs were primarily localized to the cytoplasmic side of membrane (adjusted p < 0.001) and cytoplasmic side of plasma membrane (adjusted p < 0.001) (Figure E). Biological process analysis indicated enrichment in porphyrin-containing compound biosynthetic process (adjusted p = 0.0023) and tetrapyrrole biosynthetic process (adjusted p = 0.0023), suggesting potential involvement in oxygen-related metabolic pathways (Figure F). KEGG pathway enrichment analysis revealed associations with proximal tubule bicarbonate reclamation (adjusted p = 0.2077), prion disease (adjusted p = 0.2077), adherens junction (adjusted p = 0.2077), and pathways of neurodegeneration (adjusted p = 0.2929) (Figure G). Although the enrichment levels were moderate, these results indicate potential multisystem effects of CIH.

Behavioral Assessment in CIH Mice

Behavior results showed that CIH mice exhibited reduced total distance traveled in the OFT compare to those in the normoxia (p < 0.05) (Figure A), suggesting decreased spontaneous activity and exploratory behavior. Moreover, time spent in the center of the arena was also markedly lower in the CIH mice (p < 0.05), indicating a possible increase in anxiety-like behavior (Figure B). In the Y-maze test, CIH mice showed a reduction in total arm entries (p < 0.01) (Figure D), which further indicates reduced locomotor activity, and a significant decrease in spontaneous alternation performance (SAP) (p < 0.01), indicating impairments in spatial working memory (Figure C). Together, these behavioral changes suggest alterations in locomotor activity and spatial working memory under chronic intermittent hypoxia.

4.

Behavioral outcomes in CIH and normoxia mice (n = 7 per group). (A) Total distance traveled in the open field test (OFT); (B) Time spent in the center of the arena during OFT ; (C) Spontaneous alternation performance (SAP) in the Y-maze ; (D) Total arm entries in the Y-maze . CIH: chronic intetmittent hypoxia; Nor: normoxia; *p<0.05; **p<0.01.

Shared Proteins Across Human and Mouse Proteomics

The overlapping of DEPs between the clinical human data and the CIH mouse model showed that FN1 and JCHAIN were both consistently and statistically upregulated. In the human samples, FN1 levels were positively associated with the AHI (p < 0.01) and negatively associated with the lowest oxygen saturation (p < 0.001). JCHAIN showed the same pattern, with higher expression linked to increased AHI and ESS scores, as well as lower oxygen levels.

Discussion

In this study, we combined clinical proteomics with a chronic intermittent hypoxia mouse model to explore the protein markers potentially related to neurobehavioral vulnerability in OSAHS. CIH reproduces intermittent hypoxia but does not capture key features of obstructive events or sleep fragmentation in OSAHS. , Cross-species analysis highlighted FN1 and JCHAIN as consistently upregulated proteins, associated with disease severity indices and behavioral alterations, supporting their potential as candidate markers in a hypothesis-generating cross-species framework.

Clinical Characterization and Oral Proteomics in OSAHS

OSAHS is a highly heterogeneous disorder, and previous studies have shown that different clinical subtypes may be associated with distinct comorbidity susceptibilities. A large multiethnic study from the Sleep Apnea Global Interdisciplinary Consortium (SAGIC), which included 972 patients with moderate to severe OSAHS (AHI ≥ 15 events/hour), identified distinct clinical subtypes with differing comorbidity profiles. For instance, the “disturbed sleep” subtype was associated with increased risks of hypertension, diabetes, and cardiovascular disease, while the “minimally symptomatic” subtype, despite exhibiting relatively low ESS scores, still demonstrated a heightened burden of comorbidities. These findings suggest that the clinical consequences of OSAHS may not be adequately captured by conventional metrics such as AHI alone. In our study, the OSAHS group exhibited significantly higher ESS scores than controls, confirming that our sample is enriched for the “Excessively sleepy” phenotype.

In this context, cognitive dysfunction may represent a specific and under-recognized OSAHS subtype that remains difficult to identify using current clinical tools. To address this gap, we sought to investigate noninvasive protein biomarkers through integrative proteomic analyses, aiming to improve early identification and biological characterization of cognitively vulnerable individuals within the broader OSAHS population.

CIH Model and Neurobehavioral Outcomes

In this study, clinical diagnosis of OSAHS was confirmed by PSG, while the ESS was used to characterize subjective daytime sleepiness and to enrich a clinically relevant sleepy phenotype. This clinical observation of variable cognitive outcomes in OSAHS patients prompted the need for an experimental model to further validate the behavioral manifestations associated with intermittent hypoxia. Accordingly, we established a CIH mouse model to mimic OSAHS-related hypoxic stress, while behavioral performance was assessed using the OFT and Y-maze. After 12 weeks of CIH exposure, mice displayed significantly lower body weight compared with controls, a finding of relevance given that the modeling period overlapped with growth and development, suggesting that CIH may adversely affect developmental processes. Beyond weight changes, CIH mice exhibited a marked reduction in total movement distance in the OFT, reflecting diminished spontaneous activity and behavioral motivation, which may be compatible with reduced arousal or fatigue-like behavior. In the Y-maze, a significant decline in SAP was also observed, indicating impairment in working memory and attention. These results support the use of the CIH model to explore hypoxia-related neurobehavioral changes relevant to OSAHS, while recognizing that CIH does not reproduce recurrent upper airway obstruction, intrathoracic pressure swings, or sleep fragmentation.

Cross-Species Convergence and Hypoxia-Associated Signature

At the proteomic level, 225 significantly DEPs were identified in the gingival crevicular fluid of OSAHS patients (122 upregulated, 103 downregulated). GO annotation revealed that these DEPs were enriched in peptidase inhibitor activity, complement activation, and humoral immune responses. KEGG pathway analysis further indicated their involvement in complement and coagulation cascades, suggesting a marked immune-inflammatory dysregulation in the oral protein profile of OSAHS. Correlation analysis revealed that several proteins, such as S100A8, SERPINA7, and APOA1, were associated with key clinical indicators like AHI and ESS, suggesting their relevance to disease progression.

In the oral proteomic analysis of CIH mice, 105 DEPs were identified (57 upregulated, 48 downregulated), and PCA demonstrated a clear separation between CIH and control groups. Although the specific GO/KEGG enrichment terms were not completely overlapping between humans and mice, both showed enrichment in pathways related to humoral immune responses and inflammation, indicating biological consistency between the CIH model and the immune-inflammatory processes of OSAHS.

Notably, two significantly upregulated proteins, FN1 and JCHAIN, were consistently identified in both human OSAHS patients and CIH mice. These proteins, which are centrally involved in immune-inflammatory pathways, align with the enriched GO and KEGG terms such as “humoral immune response,” and “complement activation”. These findings suggest that both proteins may reflect physiological stress related to sleep-disordered breathing. Their presence in both clinical and animal data, along with their correlations with multiple clinical features, point to their value as potential markers for identifying OSAHS patients at higher risk for cognitive problems, especially those not clearly identified by standard severity measures.

JCHAIN encodes the joining chain of polymeric IgA and IgM, an essential structural component in the formation of secretory immunoglobulins, particularly important in mucosal immune defense. , Previous studies have shown that JCHAIN deficiency leads to a shift from polymeric to monomeric IgA, reducing its epithelial transport capacity and compromising the intestinal mucosal barrier. Disruption of this barrier may affect cognitive function via the gut-brain axis. Microbiome studies suggest that overgrowth of opportunistic pathogens like Bilophila can activate Th1 immune responses, contributing to cognitive impairment. These findings align with the immune-related pathways identified in our GO/KEGG analysis, further supporting inflammation as a potential mechanism in OSA-related cognitive dysfunction.

FN1 (Fibronectin 1) is a major structural protein of the extracellular matrix (ECM), involved in cell adhesion, signal transduction, and ECM remodeling. Multiple systematic studies have established FN1 as a core player in cognitive disorders. , Large-scale GWAS have shown that FN1 gene polymorphisms, such as rs140926439, significantly reduce AD risk in APOEε4 carriers. At the pathological level, FN1 expression is elevated in brain tissues of AD patients with APOEε4, accompanied by increased gliosis. Zebrafish studies further revealed that knockout of FN1b alleviated amyloid-associated neurotoxicity. Together with the ECM-related pathways highlighted by GO/KEGG analysis in this study, FN1 appears to be not only a potential biomarker for OSAHS-related cognitive dysfunction but also a molecule with direct pathological significancemaking it a promising target for future basic research and clinical intervention.

Strengths, Limitations, and Future Directions

This study presents several notable strengths. First, it is grounded in clinical observation, specifically the finding that only a subset of OSAHS patients exhibit cognitive vulnerability, which directly informed the research question and experimental design. Second, by integrating gold-standard PSG-based diagnosis with symptom-based assessment (e.g., ESS) and combining oral proteomic data from both patients and CIH mouse models, the study establishes a robust, cross-species, multilayered framework. This design enhances the biological plausibility and internal coherence of the findings. Third, the use of oral fluids, both noninvasive and easily accessible, offers practical potential for early risk stratification, disease monitoring, and future clinical translation.

This study has several limitations. First, the clinical sample size was small and limited to a single ethnic background, so the findings should be interpreted as hypothesis generating and require validation in larger and more diverse cohorts. Second, the clinical phenotyping relied on polysomnography-defined OSAHS and subjective daytime sleepiness, and formal neuropsychological testing was not included to define cognitive phenotypes. Third, the cross-species design relies on a chronic intermittent hypoxia model that reproduces intermittent hypoxia but not obstructive events, intrathoracic pressure swings, or sleep fragmentation, and the oral fluid matrices and collection procedures differed between humans and mice, which may lead to variation in protein content. Finally, human samples were collected in the morning, whereas mouse samples were collected in the late light phase close to lights off, which corresponds to the onset of the active period in mice. Although sampling time was kept consistent within each species, it could not be fully harmonized across species because of differences in circadian and activity patterns, which may influence the oral proteome and introduce systematic bias. In addition, the present analyses are associative, so functional and mechanistic studies are needed to clarify the biological roles of the identified proteins.

Conclusion

Using 4D-DIA proteomics, this study first identified significant alterations in immune- and inflammation-related protein expression between OSAHS patients and non-OSAHS controls. These findings were further examined in a CIH mouse model, where proteomic profiling of oral protein samples revealed consistent upregulation of two key proteins, FN1 and JCHAIN. The cross-species consistency highlights these proteins as high-priority candidates with strong potential for mechanistic investigation and biomarker development in OSAHS-related cognitive vulnerability.

The mass spectrometry proteomics data have been deposited to the ProteomeXchange Consortium via the iProX partner repository with the data set identifier PXD068807.

Yuhan Xu: Conceptualization, Data curation, Formal analysis, Investigation, Methodology, Resources, Software, Writing – original draft. Mengqi Feng: Data curation, Investigation, Resources, Writing – original draft. Min Yu; Ying Xu; Xinyu Fu; Yuke Chen: Formal analysis, Investigation, Software, Writing – original draft. Chunyan Liu: Resources, Funding acquisition, Supervision, Validation, Writing – review & editing. Xuemei Gao: Conceptualization, Resources, Funding acquisition, Supervision, Validation, Writing – review & editing.

The study was supported by the National Natural Science Foundation of China (grant no. 82570125) and the Noncommunicable Chronic Diseases–National Science and Technology Major Project (grant no. 2024ZD0529100).

The authors declare no competing financial interest.