Robust and Sensitive LC-MS/MS Based Plasma Lipid Profiling on a Thermo Scientific Q Exactive HF-X Mass Spectrometer

Significance of the topic

The comprehensive and reproducible profiling of plasma lipids is essential for understanding metabolic regulation, discovering biomarkers, and supporting clinical and translational research. High-quality lipidomics requires optimized sample extraction, chromatographic separation, ionization and mass-spectrometric detection to achieve deep coverage across lipid classes while retaining sensitivity for low-abundance species.

Objectives and study overview

This study describes the development and optimization of a robust LC-MS/MS workflow for untargeted plasma lipid profiling on a Thermo Scientific Q Exactive HF-X Orbitrap mass spectrometer. The goals were to maximize lipid coverage and detection sensitivity, define optimal sample-extraction parameters, tune source and transfer conditions to minimize in-source fragmentation, and establish limits of quantification (LOQs) for isotopically labeled standards (SPLASH Lipidomix) spiked into plasma.

Methodology and sample preparation

• Biological matrix: Pooled human plasma.
• Internal standards: SPLASH Lipidomix isotopically labeled standard mixture spiked prior to extraction.
• Extraction: Modified Bligh–Dyer style using chloroform/methanol (2:1), drying and reconstitution in an organic solvent mix (ch:MeOH:PrOH 1:2:4) containing formic acid. Samples were prepared from varied plasma volumes (0–80 µL) to determine optimal input.
• Chromatography: Reversed-phase LC on a C30 stationary phase (Thermo Accucore C30, 2.1 × 150 mm, 2.6 µm) at 50 °C. Mobile phases: A = acetonitrile:water (50:50) with 5 mM formic acid; B = 2-propanol:acetonitrile:water (85:10:5) with 5 mM formic acid. Optimized gradient and a low-flow regime were used to separate diverse lipid classes.
• Mass spectrometry acquisition: Q Exactive HF-X operated in polarity switching. High-resolution MS1 survey scans at 120,000 (FWHM, 200 m/z) followed by data-dependent MS/MS at 15,000 (FWHM). Up to 10–15 MS2 events per cycle with 1.0 m/z isolation width, stepped NCE (25, 30), MS AGC target 1E6 and MS/MS AGC 1E5, max injection time ~60 ms, dynamic exclusion ~8 s.
• Source and ion-transfer optimization: Heated electrospray (HESI) optimization included sheath gas 35, aux gas 10, spray voltages ~3900 V (pos) / 3500 V (neg), capillary temperature 250 °C, heater 350 °C, and ion funnel RF tuned. RF = 35 was selected as best compromise reducing in-source fragmentation of labile lipids (cholesterol esters, ceramides) while preserving signal.
• Data processing: LipidSearch 4.2 for feature detection, product-ion driven identification and relative quantification. Identification tolerances were 5 ppm for precursor and fragment masses. Twenty-one lipid classes were included, and class-specific adduct rules were applied (e.g., PCs/LPCs as [M+H]+, TGs/MGs as [M+NH4]+, exclusion of sodiated PEs for IDs where specified).

Used instrumentation

• Thermo Scientific Q Exactive HF-X hybrid quadrupole-Orbitrap mass spectrometer.
• Thermo Scientific Accucore C30 LC column (2.1 × 150 mm, 2.6 µm) with HPLC system capable of gradient delivery and column temperature control.
• SPLASH Lipidomix isotopic standard (Avanti Polar Lipids) for monitoring extraction recovery, linearity and LOQs.

Main results and discussion

• Optimal plasma input: Extraction from 80 µL of plasma gave the best overall recovery and detection across lipid abundance ranges. Replicate extractions at this volume produced reproducible results with coefficients of variation (CVs) typically <25% for the SPLASH standards and many <10% for select low-abundance species.
• Coverage and reproducibility: The optimized workflow yielded confident identification of 577 lipid species across 13 lipid classes and quantitative data for >500 lipids. Specifically, 515 lipid species were quantified with CVs below 30% across replicate extractions/injections, indicating strong reproducibility for broad lipidome surveys.
• Sensitivity and LOQs: Limits of quantification for SPLASH standards spiked into plasma showed low ng/mL sensitivities for many lipid classes. Observed LOQs included values as low as 3 ng/mL for PE, PG and LPC standards; other examples: TG ~6 ng/mL, PC ~16 ng/mL, SM ~16 ng/mL, DG ~24 ng/mL, while PS and some sterol esters showed higher LOQs (PS ~269 ng/mL, ChoE ~182 ng/mL). Calibration curves demonstrated excellent linearity (example PE R2 ~0.9995 with 1/x weighting).
• Ion-transfer tuning: Systematic variation of ion funnel RF demonstrated a trade-off between signal-to-noise and in-source fragmentation. RF = 35 minimized fragmentation of labile lipids (e.g., ceramides, cholesterol esters) without substantially reducing signal for other classes.
• MS/MS acquisition strategy: High-resolution MS1 combined with DDA MS2 at 15k enabled confident product-ion based identification (5 ppm tolerances) and structural assignment rules per lipid class. Proper adduct selection and exclusion rules improved identification specificity.

Benefits and practical applications

• The workflow achieves deep, reproducible coverage of the plasma lipidome from modest sample volumes (80 µL), making it compatible with studies where sample is limited.
• Low LOQs for many lipid classes allow detection of low-abundance species relevant to biomarker discovery and metabolic phenotyping.
• Robust source and ion-transfer optimization reduce in-source fragmentation, improving fidelity of identifications for labile lipids.
• The approach is suitable for medium-to-large cohort studies where throughput and reproducibility are required; data-dependent acquisition enables structural annotation useful for downstream biological interpretation.

Future trends and possibilities for application

• Throughput and quantitation: Combining this optimized workflow with automated sample handling and targeted parallel reaction monitoring or data-independent acquisition could increase throughput and absolute-quantitation capability.
• Broader coverage: Integration of complementary separation strategies (e.g., HILIC, ion mobility) and expansion of isotopically labeled standard panels can improve class-specific quantitation and isomer resolution.
• Data analysis advances: Machine-learning–driven feature filtering, improved spectral libraries and standardized reporting will enhance cross-study comparisons and biomarker validation.
• Clinical translation: With further validation and harmonization, sensitive workflows like this can support clinical lipidomics applications, longitudinal studies, and multi-center trials.

Conclusion

The presented LC-MS/MS workflow on the Q Exactive HF‑X establishes a sensitive and reproducible platform for deep plasma lipid profiling. Key optimizations—80 µL plasma extraction, HESI and ion-transfer tuning, high-resolution MS1 plus DDA MS2 at 15k, and targeted data processing rules—enabled confident identification of ~577 lipids and quantitation of >500 species with acceptable precision. LOQs in the low ng/mL range for multiple lipid classes demonstrate the method’s suitability for discovery research and larger cohort studies.