基于全基因组测序技术筛选儿童金黄色葡萄球菌的致病相关分子标志物

Screening of pathogenic molecular markers of Staphylococcus aureus in children based on whole genome sequencing technology

摘要:

目的 探究儿童金黄色葡萄球菌（以下简称“金葡菌”）的分子特征，并比较不同类型菌株（感染菌株与定植菌株）的分子特征差异性，从而揭示金葡菌致病相关分子标志物。方法 采用横断面研究设计，对社区健康儿童进行鼻咽拭子采样和医院感染儿童进行临床标本采样。采用全基因组测序技术检测金葡菌耐药基因和毒素基因。采用随机森林方法筛选金葡菌致病相关标志物。结果 共检出512株金葡菌（272株感染菌株、240株定植菌株）。对于毒素基因，感染菌株组的肠毒素基因（seb和sep）、胞外酶编码基因（splA、splB、splE和edinC）、杀白细胞毒素基因（lukD、lukE、lukF-PV和lukS-PV）和表皮剥脱毒素基因（eta和etb）携带率均高于定植菌株组（均P<0.05），但肠毒素基因（sec、sec3、seg、seh、sei、sel、sem、sen、seo和seu）携带率低于定植菌株组（均P<0.05）。对于耐药基因，感染菌株组的lnuA、lnuG、aadD、tetK和dfrG耐药基因携带率明显高于定植菌株组（均P<0.05）。随机森林模型筛选金葡菌致病相关标志物，筛选前后模型交叉验证的正确率分别为69%、68%，曲线下面积分别为0.75、0.70；最终筛选出16个致病相关标志物［sem、etb、splE、sep、ser、mecA、lnuA、sea、blaZ、cat(pC233)、blaTEm-1A、aph(3')-Ⅲ、ermB、ermA、ant(9)-Ⅰa和ant(6)-Ⅰa］；变量重要性排序中前5个最重要变量为sem（OR=0.40）、etb（OR=3.95）、splE（OR=1.68）、sep（OR=3.97）、ser（OR=1.68）。结论 随机森林模型可筛选出金葡菌致病相关标志物且模型预测效果较优，为追溯高致病性金葡菌和开展精准的靶向干预提供遗传学证据。

Abstract:

Objective To explore the molecular characteristics of Staphylococcus aureus (S. aureus) in children, and to compare the molecular characteristics of different types of strains (infection and colonization strains) so as to reveal pathogenic molecular markers of S. aureus.Methods A cross-sectional study design was used to conduct nasopharyngeal swab sampling from healthy children in the community and clinical samples from infected children in the hospital. Whole genome sequencing was used to detect antibiotic resistance genes and virulence genes. A random forest method to used to screen pathogenic markers.Results A total of 512 S. aureus strains were detected, including 272 infection strains and 240 colonization strains. For virulence genes, the carrying rates of enterotoxin genes (seb and sep), extracellular enzyme coding genes (splA, splB, splE and edinC), leukocytotoxin genes (lukD, lukE, lukF-PV and lukS-PV) and epidermal exfoliating genes (eta and etb) in infection strains were higher than those in colonization strains. But the carrying rates of enterotoxin genes (sec, sec3, seg, seh, sei, sel, sem, sen, seo and seu) were lower in infection strains than in colonization strains (P<0.05). For antibiotic resistance genes, the carrying rates of lnuA, lnuG, aadD, tetK and dfrG were significantly higher in infection strains than in colonization strains (P<0.05). The accuracy of cross-validation of the random forest model for screening pathogenic markers of S. aureus before and after screening was 69% and 68%, respectively, and the area under the curve was 0.75 and 0.70, respectively. The random forest model finally screened out 16 pathogenic markers (sem, etb, splE, sep, ser, mecA, lnuA, sea, blaZ, cat(pC233), blaTEm-1A, aph(3')-III, ermB, ermA, ant(9)-Ia and ant(6)-Ia). The top five variables in the variable importance ranking were sem (OR=0.40), etb (OR=3.95), splE (OR=1.68), sep (OR=3.97), and ser (OR=1.68).Conclusions The random forest model can screen out pathogenic markers of S. aureus and exhibits a superior predictive performance, providing genetic evidence for tracing highly pathogenic S. aureus and conducting precise targeted interventions.