Surface Enhanced Raman Spectroscopy and Machine Learning for Identification of Beta-Lactam Antibiotics Resistance Gene Fragment in Bacterial Plasmid

Background: The appearance of antibiotic-resistant bacteria represents a critical medical problem with high risk to patient health. Therefore, simple, express, and reliable methods of antibiotic resistance detection should be developed.

Results: In this work, we propose a combination of highly sensitive surface-enhanced Raman spectroscopy (SERS) and machine learning (ML) for the detection of characteristic gene fragments responsible for antibiotic resistance appearance and spreading. To make the detection procedure close to the real case, we used bacterial plasmids as starting biological objects, containing or not the characteristic gene fragment (up to 1:10 ratio), encoding beta-lactam antibiotics resistance. The plasmids were subjected to enzymatic digestion and the created fragments were captured by functional SERS substrates without preliminary (bio)samples separation or purification. Based on subsequent SERS measurements, a database was created for the training and validation of ML.

Significance: The reliability of the proposed method was tested on control samples and we showed the possibility of express SEPS-ML detection of bacterial plasmids containing a characteristic gene up to the 10-7 concentration of the initial plasmid, despite the complex composition of the biological sample (i.e. the presence of the excess of alternative plasmids or various biomolecules). The proposed approach provides a good alternative to modern methods for monitoring antibiotic-resistant bacteria and is favored by its simplicity, low detection limit, and the possibility of express and unpretentious analysis.

背景：抗生素耐药菌的出现已成为严重威胁患者健康的重大医学难题，因此亟需开发简便、快速且可靠的抗生素耐药性检测方法。

结果：本研究提出将高灵敏度表面增强拉曼光谱（surface-enhanced Raman spectroscopy, SERS）与机器学习（machine learning, ML）相结合，用于检测与抗生素耐药性产生及传播相关的特征基因片段。为使检测流程更贴近实际应用场景，本研究以细菌质粒作为起始生物样本，部分质粒携带编码β-内酰胺类抗生素耐药性的特征基因片段，二者比例最高可达1:10。对质粒进行酶切处理后，所得片段无需预先进行（生物）样品分离或纯化，即可通过功能性SERS基底捕获。基于后续的SERS检测数据，构建了用于机器学习模型训练与验证的数据库。

意义：本研究通过对照样本验证了所提方法的可靠性，结果表明，即便生物样本组成复杂（如存在过量的非目标质粒或多种生物分子），仍可实现携带特征基因片段的细菌质粒的快速SERS-ML检测，初始质粒浓度低至10^-7水平时仍可检出。该方法为当前耐药菌监测技术提供了优质替代方案，其兼具操作简便、检出限低的优势，且可实现快速、简易的分析检测。