Metabolic subgroups of Diversity Outbred mice show distinct phenotypic and transcriptomic signatures of obesity and insulin resistance

Clinicians and researchers are turning towards precision medicine to treat and prevent obesity and diabetes, given the known contributions of genetics to these metabolic diseases and the wide variability reported in response to treatments. Animal models that incorporate the genetic diversity present in the human population may help discover novel genetic contributors to metabolic disease and test potential treatments. We characterized the Diversity Outbred (DO) mouse population as a model in which to study interindividual variability in metabolic disease and investigated the presence of metabolic subgroups within the population. Glucose metabolism was assessed in male Diversity Outbred (DO) mice after consumption of a high-fat diet for 14 weeks and profiled transcriptomic changes in liver, adipose, and muscle—key tissues involved in glucose homeostasis. To identify metabolic subgroups, we applied classification and regression tree analyses to metabolic phenotype measures as well as transcriptomic data. These findings suggest that DO mice exhibit a diversity of metabolic phenotypes that can be segmented into subgroups using a machine learning approach. The metabolic subgroups observed in the DO may be a useful for probing the phenotypic variability in metabolic disease observed in humans. Overall design: Male DO mice received either control or high-fat diet (HFD) for 14 weeks. At study start animals were randomized and half of the mice (N=75) were placed on the High Fat diet (60 kcal %, D12492/D12492N; Research Diets, New Brunswick, NJ) while the other half (N=75) continued receiving control diet ad libitum. Animals were examined for morbidity and clinical observation twice daily on weekdays and once daily on weekends and holidays. Body weight was measured prior to blood collection on week 1 and then weekly for the remainder of the study. Feed consumption was monitored weekly. During Week 1, 5, 9, and 14, mice were fasted for a period of 4-5 hours and resting-state glucose levels were measured. After 12 weeks, an oral glucose tolerance test was also conducted after mice were fasted overnight (approximately 16-18 hours). Serum insulin and leptin content as well as routine clinical chemistry were also measured. After 14 weeks animals were necropsied and tissues collected. Samples of liver, muscle, and adipose (abdominal) tissue were frozen in liquid nitrogen and stored at or below -70 degrees C. Total RNA was extracted from small sub-samples of these three tissues and gene expression was measured to investigate potential molecular changes driving variability among mice to allow subgroupings. Metabolic subgroups were identified using a machine-learning algorithm.

鉴于遗传学对这类代谢性疾病的明确作用，以及已有研究报道的治疗反应广泛个体差异，临床医生与研究人员正转向精准医学，以实现肥胖与糖尿病的治疗与预防。纳入人类群体固有遗传多样性的动物模型，或可助力发现代谢疾病的新型遗传致病因素，并验证潜在治疗方案。本研究对多样性远交（Diversity Outbred, DO）小鼠种群进行了系统性表征，将其作为研究代谢疾病个体间差异的模型，并探究了该种群内代谢亚群的分布情况。我们对饲喂14周高脂饮食后的雄性DO小鼠的葡萄糖代谢情况进行了评估，并对肝脏、脂肪组织与肌肉——这些参与葡萄糖稳态的关键组织——的转录组变化进行了谱分析。为识别代谢亚群，我们针对代谢表型测量数据与转录组数据应用了分类与回归树分析方法。本研究结果表明，DO小鼠展现出多样化的代谢表型，可通过机器学习方法划分为不同亚群。本研究在DO小鼠中观察到的代谢亚群，或可用于探究人类代谢疾病中存在的表型多样性。
实验整体设计：雄性DO小鼠分别接受正常饮食或高脂饮食（High-Fat Diet, HFD）干预，干预周期为14周。实验初始阶段对小鼠进行随机分组，其中75只小鼠饲喂高脂饲料（热量占比60%，型号D12492/D12492N；购自美国新泽西州纽布伦斯威克市Research Diets公司），剩余75只小鼠则自由采食正常饲料。实验期间，工作日每日对小鼠进行两次发病率监测与临床状态观察，周末及节假日每日观察一次。于第1周采血前测量小鼠体重，后续每周测量一次直至实验结束。每周监测饲料消耗量。在第1、5、9及14周，小鼠禁食4-5小时后测量其静息血糖水平。实验第12周，小鼠经过夜禁食（约16-18小时）后接受口服葡萄糖耐量试验。同时检测血清胰岛素与瘦素含量，并开展常规临床生化指标检测。14周实验周期结束后，对小鼠实施安乐死并采集组织样本。将肝脏、肌肉与腹部脂肪组织样本置于液氮中速冻，随后保存于-70℃及以下环境。从这三种组织的少量子样本中提取总RNA，检测基因表达水平，以探究驱动小鼠个体间代谢差异的潜在分子变化，进而实现亚群划分。本研究通过机器学习算法完成代谢亚群的识别。