Introduction: Multidimensional measurements such as the integration of liquid chromatography, ion mobility spectrometry and mass spectrometry (LC-IMS-MS) have been implemented to separate lipid species and provide valuable polarity, structural and mass information simultaneously. LC-IMS-MS measurements enable improved lipid characterization and identification but result in large and complex datasets which are extremely difficult to process. Thus, developing software capable of accurate and rapid molecular analyses is essential.
Objectives: The primary objective of this study was to modify the free, open source software Skyline for rapid and confident annotation of lipidomic data using manually curated sample-specific lipid spectral libraries containing LC, IMS, MS and MS/MS information. The secondary objective was to apply our library to assess lipids in bronchoalveolar lavage fluid (BALF) and plasma from patients who had various outcomes following bronchoscopy treatment for severe smoke inhalation.
Methods: To create the lipid spectral libraries, hundreds of target lipids from multiple lipid categories were analyzed in both positive and negative ionization modes. Each target lipid was populated with an m/z, LC retention time, IMS collision cross section (CCS) and fragments. These values were manually extracted from various sample types and verified using existing literature. A set of ~20 endogenous lipids for each sample type were set as the iRT normalization lipids and used to allow gradient time changes and LC alignment. To utilize the lipid libraries with clinical samples, lipidomic assays were performed on 21 blood plasma samples and 99 BALF samples from patients with severe smoke inhalation using the LC-IMS-MS/MS platform in positive and negative ionization modes. The lipids were identified using our lipid spectral libraries in Skyline and evaluated using statistical tests.
Results: The multidimensional lipid spectral libraries in Skyline allowed rapid and targeted processing of all lipids in the library simultaneously. The iRT calculator allowed utilization of different gradient times and was verified for use with LC retention time shifts due to run-to-run variation and differences in columns, instruments, and labs. Application of the collision cross section (CCS) filtering further increased lipid annotation confidence and greatly improved the signal to noise ratio for the target species. Applying the library to smoke inhalation patients having various outcomes following bronchoscopy treatment pinpointed lipidomic changes in the different groups. This evaluation is ongoing to fully understand the significance of the findings and determine if personalized treatment is possible based on lipid content.
Conclusion: The sample-specific lipid spectral libraries created and implemented in Skyline provides the scientific community with an essential tool for rapid and confident lipid analyses which can be applied to a variety of clinical studies.