Objective: To investigate the functional changes of key gut microbiota (GM) that produce lipopolysaccharide (LPS) in atrial fibrillation (AF) patients and to explore their potential role in the pathogenesis of AF. Methods: This was a prospective cross-sectional study. Patients with AF admitted to Beijing Chaoyang Hospital of Capital Medical University were enrolled from March 2016 to December 2018. Subjects with matched genetic backgrounds undergoing physical examination during the same period were selected as controls. Clinical baseline data and fecal samples were collected. Bacterial DNA was extracted and metagenomic sequencing was performed by using Illumina Novaseq. Based on metagenomic data, the relative abundances of KEGG Orthology (KO), enzymatic genes and species that harbored enzymatic genes were acquired. The key features were selected via the least absolute shrinkage and selection operator (LASSO) analysis. The role of GM-derived LPS biosynthetic feature in the development of AF was assessed by receiver operating characteristic (ROC) curve, partial least squares structural equation modeling (PLS-SEM) and logistic regression analysis. Results: Fifty nonvalvular AF patients (mean age: 66.0 (57.0, 71.3), 32 males(64%)) were enrolled as AF group. Fifty individuals (mean age 55.0 (50.5, 57.5), 41 males(82%)) were recruited as controls. Compared with the controls, AF patients showed a marked difference in the GM genes underlying LPS-biosynthesis, including 20 potential LPS-synthesis KO, 7 LPS-biosynthesis enzymatic genes and 89 species that were assigned as taxa harbored nine LPS-enzymatic genes. LASSO regression analysis showed that 5 KO, 3 enzymatic genes and 9 species could be selected to construct the KO, enzyme and species scoring system. Genes enriched in AF group included 2 KO (K02851 and K00972), 3 enzymatic genes (LpxH, LpxC and LpxK) and 7 species (Intestinibacter bartlettii、Ruminococcus sp. JC304、Coprococcus catus、uncultured Eubacterium sp.、Eubacterium sp. CAG:251、Anaerostipes hadrus、Dorea longicatena). ROC curve analysis revealed the predictive capacity of differential GM-derived LPS signatures to distinguish AF patients in terms of above KO, enzymatic and species scores: area under curve (AUC)=0.957, 95%CI: 0.918-0.995, AUC=0.940, 95%CI 0.889-0.991, AUC=0.972, 95%CI 0.948-0.997. PLS-SEM showed that changes in lipopolysaccharide-producing bacteria could be involved in the pathogenesis of AF. The key KO mediated 35.17% of the total effect of key bacteria on AF. After incorporating the clinical factors of AF, the KO score was positively associated with the significantly increased risk of AF (OR<0.001, 95%CI:<0.001-0.021, P<0.001). Conclusion: Microbes involved in LPS synthesis are enriched in the gut of AF patients, accompanied with up-regulated LPS synthesis function by encoding the LPS-enzymatic biosynthesis gene.
目的： 分析心房颤动（房颤）患者肠道菌群脂多糖合成功能的变化特点及其可能的作用。 方法： 本研究为前瞻性、横断面研究。选取2016年3月至2018年12月在首都医科大学附属北京朝阳医院住院的非瓣膜性房颤患者作为房颤组，并选取同期遗传背景匹配的体检者作为对照。收集所有研究对象的临床基线资料及粪便样本，提取肠道细菌DNA，采用Illumina novaseq进行宏基因组学测序。基于宏基因组学测序数据，分析脂多糖合成过程中的同源基因簇（KEGG orthology，KO）、合成酶的编码基因及含有这些酶基因的细菌的丰度。基于LASSO分析，筛选与房颤相关的参与脂多糖合成的关键因子，并采用受试者工作特征（ROC）曲线分析、中介分析和logistic回归分析评价其在房颤发病中可能的作用。 结果： 房颤组50例，年龄66.0（57.0，71.3）岁，男性32例（64%）。对照组50名，年龄55.0（50.5，57.5）岁，男性41例（82%）。有20个参与脂多糖合成的KO、7个脂多糖合成酶的编码基因和89个同时可编码9个脂多糖合成酶的肠道细菌在对照组与房颤组间存在差异。LASSO回归分析显示，有5个KO、3个酶基因及9个种层级物种被筛选为关键因子，其中富集在房颤组的因子包括：2个KO（K02851和K00972），3个酶基因（LpxH、LpxC和LpxK），7个种层级物种（Intestinibacter bartlettii、Ruminococcus sp. JC304、Coprococcus catus、uncultured Eubacterium sp.、Eubacterium sp. CAG：251、Anaerostipes hadrus、Dorea longicatena）。基于上述关键因子构建回归模型，ROC曲线分析示：KO、酶基因、物种分值识别房颤的曲线下面积（AUC）和95%CI分别为0.957（0.918~0.995）、0.940（0.889~0.991）和0.972（0.948~0.997）。中介分析的结果显示，参与脂多糖合成的肠道细菌的变化可影响房颤发病，其中有35.17%是通过富集相应的KO所介导。多因素logistic回归分析结果显示，KO分值绝对值增大，房颤的发生风险增加（OR值<0.001，95%CI：<0.001~0.021，P<0.001）。 结论： 房颤患者肠道内富集了参与脂多糖合成的细菌，其通过编码脂多糖合成酶基因，使得房颤患者脂多糖合成功能有所上调。.