Network Analysis to Identify Multi-Omic Correlations in the Lower Airways of Children With Cystic Fibrosis.
Network Analysis to Identify Multi-Omic Correlations in the Lower Airways of Children With Cystic Fibrosis.
复制标题
DOI：
10.3389/fcimb.2022.805170
复制
发表时间：
2022
影响因子：
5.7
通讯作者：
Laguna TA
中科院分区：
文献类型：
--
作者：
O'Connor JB;Mottlowitz M;Kruk ME;Mickelson A;Wagner BD;Harris JK;Wendt CH;Laguna TA
囊性纤维化（CF）发病率和死亡率的主要原因是继发于慢性气道感染和炎症的进行性肺病;然而，对CF气道感染和炎症的驱动因素尚不清楚。通过提供气道的生理快照，代谢组学可以深入了解这些过程。将代谢组学数据与微生物组数据和表型测量相关联可以揭示代谢物，下呼吸道细菌群落和疾病结果之间的复杂关系。在这项研究中，我们表征了CF（PWCF）和疾病对照（DC）受试者支气管肺泡灌洗液（BALF）样本中的气道代谢组，并使用多组学网络分析来确定与气道微生物组的相关性。Biocrates靶向液相色谱质谱（LC-MS）平台用于测量临床指征支气管镜检查期间获得的BALF中的409个代谢组学特征。使用定量聚合酶链反应（qPCR）测量总细菌负荷（TBL）。使用Qiagen EZ 1 Advanced自动提取平台提取DNA，并使用16 S测序进行细菌谱分析。使用Wilcoxon秩和检验单变量评估疾病组间代谢组学特征的差异，并使用随机森林（RF）来识别区分各组的特征。将这些特征与TBL和炎症标志物进行比较，包括白色血细胞计数（WBC）和中性粒细胞百分比。稀疏监督典型相关网络分析（SsCCNet）用于评估多组学相关性。CF代谢组的特点是增加氨基酸和减少酰基肉毒碱。氨基酸和酰基肉毒碱也是与炎症和细菌负荷最密切相关的特征之一。RF确定了CF状态的强代谢组学预测因子，包括L-蛋氨酸-S-氧化物。SsCCNet确定了代谢组和微生物组之间的相关性，包括传统CF病原体葡萄球菌（Staphylococcus）、一组非传统分类群（包括普雷沃氏菌（Prevotella））和特定代谢组学标记子网络之间的相关性。总之，我们的工作确定了CF气道特有的代谢组学特征，并揭示了值得进一步研究的多组学相关性。
The leading cause of morbidity and mortality in cystic fibrosis (CF) is progressive lung disease secondary to chronic airway infection and inflammation; however, what drives CF airway infection and inflammation is not well understood. By providing a physiological snapshot of the airway, metabolomics can provide insight into these processes. Linking metabolomic data with microbiome data and phenotypic measures can reveal complex relationships between metabolites, lower airway bacterial communities, and disease outcomes. In this study, we characterize the airway metabolome in bronchoalveolar lavage fluid (BALF) samples from persons with CF (PWCF) and disease control (DC) subjects and use multi-omic network analysis to identify correlations with the airway microbiome. The Biocrates targeted liquid chromatography mass spectrometry (LC-MS) platform was used to measure 409 metabolomic features in BALF obtained during clinically indicated bronchoscopy. Total bacterial load (TBL) was measured using quantitative polymerase chain reaction (qPCR). The Qiagen EZ1 Advanced automated extraction platform was used to extract DNA, and bacterial profiling was performed using 16S sequencing. Differences in metabolomic features across disease groups were assessed univariately using Wilcoxon rank sum tests, and Random forest (RF) was used to identify features that discriminated across the groups. Features were compared to TBL and markers of inflammation, including white blood cell count (WBC) and percent neutrophils. Sparse supervised canonical correlation network analysis (SsCCNet) was used to assess multi-omic correlations. The CF metabolome was characterized by increased amino acids and decreased acylcarnitines. Amino acids and acylcarnitines were also among the features most strongly correlated with inflammation and bacterial burden. RF identified strong metabolomic predictors of CF status, including L-methionine-S-oxide. SsCCNet identified correlations between the metabolome and the microbiome, including correlations between a traditional CF pathogen, Staphylococcus, a group of nontraditional taxa, including Prevotella, and a subnetwork of specific metabolomic markers. In conclusion, our work identified metabolomic characteristics unique to the CF airway and uncovered multi-omic correlations that merit additional study.