Exploratory analysis of sleep deprivation effects on gene expression and regional brain metabolism

Sleep deprivation affects cognitive performance and immune function, yet its mechanisms and biomarkers remain unclear. This study explored the relationships among gene expression, brain metabolism, sleep deprivation, and sex differences.

Methods

Fluorodeoxyglucose-18 positron emission tomography (18F-FDG PET) measured brain metabolism in regions of interest (ROIs), and RNA analysis of blood samples assessed gene expression pre- and post-sleep deprivation. Mixed model regression and principal component analysis (PCA) identified significant genes and regional metabolic changes.

Results

There were 23 and 28 differentially expressed probesets for the main effects of sex and sleep deprivation, respectively, and 55 probesets for their interaction (FDR-corrected p<0.05). Functional analysis revealed enrichment in nucleoplasm- and UBL conjugation-related genes. Genes showing significant sex effects mapped to chromosomal regions Y and 19 (Benjamini-Hochberg (BH) FDR p<0.05), with 11 genes (4%) and 29 genes (10.5%) involved, respectively. Differential gene expression highlighted sex-based differences in innate and adaptive immunity. 

For brain metabolism, sleep deprivation resulted in significant decreases in the left insula, medial prefrontal cortex (BA32), somatosensory cortex (BA1/2), and motor premotor cortex (BA6) and increases in the right inferior longitudinal fasciculus, primary visual cortex (BA17), amygdala, cerebellum, and bilateral pons. Hemispheric asymmetry in brain metabolism was observed, with BA6 decreases correlating with increased UBL conjugation gene expression.

Conclusion

Sleep deprivation broadly impacts brain metabolism, gene expression, and immune function, revealing cellular stress responses and hemispheric vulnerability. These findings enhance understanding of the molecular and functional effects of sleep deprivation.

Methods
Sleep Deprivation 

Eight healthy subjects, 4 male and 4 female, were recruited from the University of California Irvine, after IRB approval. On day 1, subjects were initially assigned a 24-hour period of normal activity (e.g. walk, talk, study, watch TV, play games, use the computer, etc.). These subjects were tested on the Psychomotor Vigilance Test (PVT) and asked to rate their subjective level of sleepiness on the Stanford Sleepiness Scale (SSS) at baseline. Higher scores indicate a longer, more delayed, response time on the PVT, while higher scores on the SSS indicate greater degrees of sleepiness. The SSS scale is shown in Table 1. Each subject’s performance on the Psychomotor Vigilance Test (PVT), and subjective sleepiness ratings (SSS) were recorded both before and after sleep deprivation (Table 2). There was no significant difference in age between male and female subjects (Table 3), all of whom had no prior psychiatric history. 

Blood samples were collected on baseline day at 1 p.m, pre-sleep deprivation (pre-SD). Sleep deprivation activities and blood sample acquisition times are recorded in Table 4. At the end of day 1 (11 p.m), subjects were moved to an outpatient research facility for the sleep deprivation protocol. They were requested not to nap or sleep during the sleep deprivation period, and were additionally tasked with filling out forms and answering questions about their mood every two to four hours. Staff members monitored the subjects during the sleep deprivation period. Subjects were allowed to walk, talk, study, watch TV, play games or cards, read, and use the computer, but were not allowed caffeinated foods or beverages. A second blood sample was collected 18 hours after starting sleep deprivation activities (SD Day 2, 1 p.m), subjects completed the protocol and were driven home by cab. 

Gene Data Processing

Blood samples (3 ml) were drawn from each subject, into Tempus tubes (ABI, ThermoFisher, Carlsbad, CA) 24 hours apart. The blood samples collected at baseline and 18 hours after starting sleep deprivation activities were processed with Affymetrix HG-U133 Plus 2.0 gene expression microarray chips according to the manufacturer’s instructions (Affymetrix, ThermoFisher, Carlsbad CA). Data processing was done using R 4.2 and BioConductor 3.16 [32]. The Affymetrix HG-U133 Plus 2.0 microarray ‘cel’ files were read using the affy routine with the hgu133plus2.db package. Quantile normalization was used to standardize probeset data [33]. A linear model was fitted to the expression data for each probeset using ‘lmfit’ from the limma package, to eliminate weakly expressed probesets, and the top 40,000 probesets were found using the topTables function. Type III mixed ANOVA was implemented  using the lmerTest library in R, with the main effects being sex, sleep deprivation, and sleep deprivation-sex interaction. Age and RNA integrity number (RIN) were used as covariates. The top 300 probesets for each main effect from mixed ANOVA and PCA were analyzed for enrichment using the Database for Annotation, Visualization and Integrated Discovery (DAVID) [34; 35]. Principal component analysis was conducted using the pca function with normalized and scaled expression data. 

F18-FDG PET Scan Processing

The pre-SD and post-SD F18 FDG-PET scans were obtained from each subject.  Each F18-FDG PET scan was normalized in MATLAB (Mathworks, Sherborn, Massachusetts, USA) using Statistical Parametric Mapping (SPM) 5 software (Functional Imaging Laboratory, Wellcome Department of Cognitive Neurology, University College London, London, UK) to spatially transform the images to a template conforming to the space derived from standard brains from the Montreal Neurological Institute, and convert it to the space of the stereotactic atlas of Talairach and Tournoux. The images were then smoothed with a Gaussian low-pass filter of 8mm to minimize noise and improve spatial alignment.

Regions of interest (ROI) analysis was done by extracting metabolic values from regions of interest using VINCI (“Volume Imaging in Neurological Research, Co-Registration and ROI included”) software. Supplementary Figure 1 shows ROI segmentation of FDG-PET scans labeled with brain regions and Brodmann areas (BA). 

A type III mixed two-way ANOVA was implemented using the lmerTest library in R. The model considered sex as a between-subjects factor and condition (pre-sleep deprivation vs. post-sleep deprivation) as a within-subjects factor. Principal component analysis was performed using the pca() function in the BioConductor environment [32] in R. Prior to extracting principal components, all probesets were scaled by extracting the mean value and dividing by the standard deviation for that variable in R.

睡眠剥夺会影响认知表现与免疫功能，但其背后的机制与生物标志物仍不明确。本研究探讨了基因表达、脑代谢、睡眠剥夺以及性别差异之间的关联。

方法
氟脱氧葡萄糖-18正电子发射断层扫描（Fluorodeoxyglucose-18 positron emission tomography，18F-FDG PET）用于检测感兴趣脑区（regions of interest，ROIs）的代谢水平，同时通过血液样本的RNA分析评估睡眠剥夺前后的基因表达情况。研究采用混合模型回归与主成分分析（principal component analysis，PCA）来筛选显著差异的基因与脑区代谢变化。

结果
针对性别与睡眠剥夺的主效应，分别存在23个和28个差异表达探针集；二者的交互效应则对应55个探针集（错误发现率校正后p<0.05）。功能富集分析显示，这些基因主要富集在核质类与泛素样蛋白（ubiquitin-like protein，UBL）缀合相关的通路中。存在显著性别效应的基因定位于Y染色体与19号染色体区域（Benjamini-Hochberg（BH）错误发现率校正p<0.05），分别涉及11个基因（占比4%）与29个基因（占比10.5%）。差异基因表达分析揭示了先天免疫与适应性免疫中的性别差异。
在脑代谢方面，睡眠剥夺可导致左侧岛叶、内侧前额叶皮层（Brodmann分区32区，BA32）、躯体感觉皮层（BA1/2区）以及运动前皮层（BA6区）的代谢水平显著降低，同时使右侧下纵束、初级视觉皮层（BA17区）、杏仁核、小脑以及双侧脑桥的代谢水平升高。研究观察到脑代谢存在半球不对称性，其中BA6区的代谢降低与UBL缀合基因的表达上调呈显著相关。

结论
睡眠剥夺可广泛影响脑代谢、基因表达与免疫功能，揭示了细胞应激反应与半球易感性差异。本研究结果加深了对睡眠剥夺的分子与功能效应的理解。

方法
睡眠剥夺方案
本研究从加州大学欧文分校招募了8名健康受试者（4名男性、4名女性），研究已获得伦理审查委员会（Institutional Review Board，IRB）批准。第1天，受试者先进行24小时的正常活动（如散步、交谈、学习、看电视、玩游戏、使用电脑等）。基线阶段，受试者需接受精神运动警戒测试（Psychomotor Vigilance Test，PVT），并通过斯坦福嗜睡量表（Stanford Sleepiness Scale，SSS）自评嗜睡程度。PVT得分越高代表反应时越长、延迟越明显；SSS得分越高则代表嗜睡程度越严重。SSS量表详见表1。所有受试者在睡眠剥夺前后均完成了PVT测试与SSS自评（表2）。男性与女性受试者的年龄无显著差异（表3），且所有受试者均无精神疾病史。
血液样本于基线日下午1点，即睡眠剥夺前（pre-SD）采集。睡眠剥夺期间的活动与血液样本采集时间详见表4。第1天晚11点，受试者被转移至门诊研究设施开展睡眠剥夺方案。研究人员要求受试者在睡眠剥夺期间不得小睡或入睡，并需每2至4小时填写一次情绪相关问卷与回答问题。研究人员在睡眠剥夺期间全程监测受试者状态。受试者可进行散步、交谈、学习、看电视、玩游戏或纸牌、阅读以及使用电脑，但不得摄入含咖啡因的食物与饮品。在开始睡眠剥夺活动18小时后（睡眠剥夺第2天下午1点），采集第二份血液样本。受试者完成全部实验流程后，乘坐出租车返回家中。

基因数据处理
从每位受试者体内采集3ml血液样本，置于Tempus管（ABI，赛默飞世尔科技，卡尔斯巴德，加利福尼亚州）中，两次采样间隔24小时。基线与睡眠剥夺18小时后采集的血液样本，按照制造商说明书使用Affymetrix HG-U133 Plus 2.0基因表达微阵列芯片进行处理（Affymetrix，赛默飞世尔科技，卡尔斯巴德，加利福尼亚州）。数据处理采用R 4.2与BioConductor 3.16软件[32]。使用affy程序包与hgu133plus2.db数据包读取Affymetrix HG-U133 Plus 2.0微阵列的‘cel’文件。采用分位数标准化对探针集数据进行标准化处理[33]。针对每个探针集的表达数据，使用limma包中的‘lmfit’函数构建线性模型以剔除低表达探针集，并通过topTables函数筛选出排名前40000的探针集。使用R语言的lmerTest库实施III型混合方差分析，其中主效应为性别、睡眠剥夺以及睡眠剥夺-性别交互效应，以年龄与RNA完整性数（RNA integrity number，RIN）作为协变量。将混合方差分析与主成分分析得到的每个主效应排名前300的探针集，通过数据库注释、可视化与整合发现工具（Database for Annotation, Visualization and Integrated Discovery，DAVID）[34; 35]进行富集分析。使用pca函数对标准化与缩放后的表达数据开展主成分分析。

F18-FDG PET扫描处理
采集每位受试者睡眠剥夺前与睡眠剥夺后的F18-FDG PET扫描图像。使用MATLAB（Mathworks，谢尔本，马萨诸塞州，美国）的统计参数映射5（Statistical Parametric Mapping，SPM 5）软件（伦敦大学学院威康认知神经学系功能成像实验室，伦敦，英国）对每幅F18-FDG PET扫描图像进行标准化处理，将图像空间转换至符合蒙特利尔神经学研究所标准脑的模板空间，并转换至Talairach和Tournoux立体定位图谱空间。随后使用8mm高斯低通滤波器对图像进行平滑处理，以降低噪声并优化空间对齐效果。
采用感兴趣脑区分析：使用VINCI（"神经研究体积成像、配准与感兴趣脑区分析"，"Volume Imaging in Neurological Research, Co-Registration and ROI included"）软件从感兴趣脑区中提取代谢数值。补充图1展示了标注有脑区与Brodmann分区（BA）的FDG-PET扫描图像的感兴趣脑区分割结果。
使用R语言的lmerTest库实施III型混合双向方差分析，模型中将性别作为被试间因素，将条件（睡眠剥夺前vs睡眠剥夺后）作为被试内因素。使用R语言BioConductor环境[32]中的pca()函数开展主成分分析。在提取主成分前，对所有探针集进行缩放处理：即提取每个变量的均值并除以其标准差。