Hello everyone, today I'm sharing an article published in Nature Chemistry in 2024: Diazobutanone-assisted isobaric labelling of phospholipids and sulfated glycolipids enables multiplexed quantitative lipidomics using tandem mass spectrometry. This article introduces a diazobutanone-assisted isotope labeling strategy. The diazobutanone reagent can be coupled to phosphodiester or sulfate groups and can accommodate various functional groups on different lipids, thus enabling high-throughput, multiplexed quantitative isotope labeling for multiplex quantitative lipidomics of various lipids (including various phospholipids and glycolipids).

Lipidomics is a rapidly developing field aiming to characterize the lipidome in biological systems and elucidate lipid functions and their roles in disease development. Phospholipids (PLs) and glycolipids are the main components of liposomes and play many important roles in organisms. The regulation of cellular systems by the lipidome is determined by the relative abundance of each lipid type and its chemical structure. Recently, the association between the lipidome and disease has spurred the development of quantitative lipidomics for the discovery of disease biomarkers and therapeutic targets. However, due to the high diversity of lipids in chemical structure and physicochemical properties, high-throughput quantitative lipidomics methods lag behind genomics and proteomics. Therefore, a high-throughput strategy capable of accurately identifying and quantifying a wide range of lipids is crucial for addressing key biological questions related to the lipidome and lipid regulation.

First, the authors designed a diazonium ketone compound containing a diazonium group capable of O-alkylating the phosphodiester bond, followed by a carbonyl group for subsequent isotopic labeling. To demonstrate the broad applicability of diazonium ketone to a variety of lipids, the authors tested it with nine representative phosphate-containing lipids, including phosphatidylcholine (PC), phosphatidylethanolamine (PE), phosphatidylglycerol (PG), phosphatidylserine (PS), phosphatidic acid (PA), phosphatidylinositol (PI), lysophosphatidylcholine (LysoPC), ether-phosphatidylcholine (etherPC), and sphingomyelin (SM). Under optimized reaction conditions (75 mM diazonium ketone and 0.24 mM fluoroboric acid catalyst in ethyl propionate for 35 min at room temperature), the average conversion rate of all lipid standards was 98%, with minimal side reactions.

In the second step of labeling, isotopic labeling was performed using the aminoxyTMT mass tag developed for carbonyl-containing biomolecules. The lipid to aminoxyTMT ratio was 1:2, and the reaction solvent was 10% isopropanol/methanol and 0.1% acetic acid. Near-complete labeling efficiency was achieved after 15 minutes of reaction. Excess labeling reagent could be removed by extraction with water and ethyl acetate. The entire labeling process can be completed within 2 hours, enabling rapid, economical, and efficient high-throughput quantitative experiments.

Following labeling, primary mass spectrometry showed a 372 Da mass increment introduced onto the lipids, composed of ketone and aminoxyTMT. The PA mass increased by 744 Da due to the addition of a 2-unit tag. Compared to previously reported diazomethanes, diazonyl ketones exhibit weaker reactivity and milder reaction conditions, with no observed alkylation on the amino group and lower PL fragmentation, making it less likely for PS products to be confused with PC or PA. However, to avoid inaccurate PA quantification, it is best to separate PA from other PL types during analysis.

Representative MS2 spectra of labeled PE 15:0-18:1 (d7) and PG 16:0-18:1 show that the peaks at m/z 126-131 represent reporter ions generated by the aminoxyTMT tag used for multiplex quantification. The peaks at m/z 513 and 544 represent fragments of aminoxyTMT formed by the PE and PG head groups and oximes, indicating lipid precursor ions. The peaks at 570 and 577 represent the neutral loss of polar head groups, which can be elucidated by MS3 fragmentation to distinguish the fatty acid chain and differentiate lipid isomers.

To evaluate the quantitative performance of this method, the authors labeled and analyzed five different amounts of lipid plasmas (PLs) (300 ng, 60 ng, 30 ng, 12 ng, and 3 ng). The correlation between the measured ratio of reporter ions and the expected ratio of lipids was examined three times, showing a linear relationship with an R² value of 0.9985 and a slope of 0.9960. The retention times of PLs labeled with different channels did not change significantly, ensuring simultaneous MS² analysis of PLs with different labeled variants. The authors also evaluated the sensitivity of the method by comparing the limits of quantitation (LOQ) before and after labeling with lipid standards. The results showed that the LQ generated by labeling was comparable to that of unlabeled lipids, indicating that the method successfully facilitated multiplex quantification without affecting sensitivity.

Subsequently, to demonstrate the applicability of this method to complex biological samples, the authors applied it to analyze major PLs (PC, PE, PI, PS, PG, and PA) in the liver tissue of the human pancreatic cancer cell line PANC-1 and obese mice. The two-step isotope labeling and mass spectrometry workflow assisted by diazobutanone is shown in the figure below. Deuterated lipids were added to the liver tissue homogenate before homogenization for lipid mass estimation and correction of lipid extraction recovery. After Folch lipid extraction, six samples were hexadiazobutanone-assisted isotope labeling and pooled before LC-MS/MS analysis.

The labeled mixtures were analyzed by LC-MS/MS in positive ion mode. Labeled PLs were identified by precise mass matching with the database and by classifying diagnostic ions using an internally developed R script. Quantification was obtained by extracting the abundance of reporter ions from the six aminoxyTMT channels. All MS2 fragmentation patterns, charges, and retention times were considered to validate their identification. Using the improved LipidBlast database, the authors identified a total of 304 PLs in PANC-1 cells and 198 PLs in mouse liver tissue. In mouse liver tissue, PC accounted for 43% of the identified PLs, PE for 29%, and PG for 13%. The lipid composition was comparable to previous label-free lipidomics analyses of mouse livers with non-alcoholic steatohepatitis. These results demonstrate the high reproducibility and reliable lipidomics capabilities of the new method.

To assess the quantitative accuracy of this method on complex samples, the authors performed hexameric diazobutanone-assisted isotope labeling on six aliquots of PL extract. Diazobutanone reaction and hexameric aminoxyTMT labeling were performed separately. Mass spectrometry analysis was performed after mixing the mixtures at molar ratios of 1:1:1:1:1:1 and 1:1:2:4:6:8, respectively. Furthermore, deuterated PL standards were added to the mixtures in a 1:2:4:4:2:1 ratio to further assess quantitative accuracy. The proportion of the reporting ion in the deuterated PL standards was consistent with expectations, without proportional distortion. At theoretical ratios of 1:1:1:1:1:1 and 1:1:2:4:6:8, the median coefficient of variation (CV) was less than 10%, and the relative standard deviation was less than 0.09, demonstrating good quantitative performance.

The authors then performed a sixfold quantitative analysis on liver tissues from three healthy, lean male mice and three insulin-resistant, obese male (ob/ob) mice using the same workflow. A total of 251 phospholipids (PLs) with acyl chains were identified in the liver tissues of both lean and obese mice. First, the authors examined the PC/PE ratio, a progression indicator of liver disease, using the reporter ion intensity from the PLs. They found a slight decrease in this ratio in obese mice, possibly due to insulin resistance, but not yet the development of severe liver disease. The authors found that phospholipids with longer acyl chains (>38 carbon atoms) were upregulated in obese mice, while PLs with shorter acyl chains (<34 carbon atoms) and highly unsaturated PCs were reduced. This trend is consistent with previous reports and can be attributed to the elongation of long-chain fatty acids catalyzed by the elongation enzyme Elov16, which has been reported to play a key role in obesity-induced insulin resistance. A similar pattern has also been observed in mice with liver disease, suggesting that phospholipid chain elongation in obesity may affect liver function.

This article reports a novel isotope-labeled multiplex quantitative lipidomics approach that can be applied to study multiple types of lipids and their dynamic interactions in various diseases, and facilitate the non-targeted discovery of lipid biomarkers for different diseases and physiological states. The diazonium ketone-assisted isotope labeling strategy provides a strong starting point for lipidomics, including higher quantitative accuracy, reproducibility, fewer missing values, elimination of matrix effects, and the ability to analyze replicates or multiple test sets in the same experiment. In conclusion, diazonium ketone-assisted isotope labeling will propel the field of quantitative lipidomics to unprecedented high-throughput analysis with greater coverage and higher sensitivity.