A high throughput lipidomics method using scheduled multiple reaction monitoring

Lipid compositions of cells, tissues and bio-fluids are complex, with varying concentrations and structural diversity, which makes their identification challenging. Newer methods for comprehensive analysis of lipids are thus necessary. Herein, we propose a targeted-mass spectrometry based method for large-scale lipidomics using a combination of variable retention time window and relative dwell time weightage. Using this, we detected more than 1000 lipid species, including structural isomers. The limit of detection varied from femtomolar to nanomolar range and the coefficient of variance <30% for 849 lipid species. We used this method to identify lipids altered due to Vitamin B12 deficiency and found that the levels of lipids with ω-3 fatty acid chains decreased while those with ω-6 increased. This method enables identification of by far the largest number of lipid species with structural isomers in a single experiment and would significantly advance our understanding of the role of lipids in biological processes.

improved by varying dwell time weightage for each transition without compromising with 104 the cycle time (https:// https://sciex.com/). Assigning a low dwell time weightage to high 105 abundant compounds and high dwell time weightage to less abundant compounds, 106 irrespective of the elution window, may help in accommodating large number of 107 transitions in a single run with improved data quality. 108 Leveraging the combinatorial optimization of scheduled-MRM, variable RT window and 109 dwell time weightage, we report a rapid and sensitive targeted lipidomics method 110 capable of identifying more than 1000 lipid species, including isomers of triglycerides, 111 diglycerides, and phospholipids in a single MS run-time of 24 minutes. To the best of 112 our knowledge, this is the largest number of lipid species identified till date in a single 113 experiment. 114 We further, exploited this method to quantitate isomer specific different lipid classes in 115 vitamin B 12 deficiency in the context of Indian population. Previously, we have shown 116 that vitamin B12 deficiency alters the lipid metabolism to drive cardiometabolic 117 phenotype in rats 39 . This study clearly demonstrates the effects of vitamin B 12 deficiency 118 with changes in the specific lipidomic isomers, laying the foundation to understand the 119 development of highly prevalent cardio-metabolic diseases in a strictly vegetarian diet 120 adhered country like India.

121
Results: 122 We developed a scheduled-MRM method that can identify more than 1000 lipid species 123 in a single mass spectrometric acquisition using a combination of variable-RTW and   145 In an attempt to identify different lipid isomers, we used customized-approaches for   Limit of blank (LoB), limit of detection (LoD), limit of quantitation (LoQ), and linear 186 range.

187
The raw analytical signal in blank was considered for establishing the LoB, which was  (table 1).

206
To determine the percent recovery of all the lipid species, a known amount of lipid 207 standards were added to plasma (matrix) before or after (spike) extraction of the lipids 208 from the plasma. The raw area signals obtained from these two conditions were 209 compared to determine the percentage recovery. These experiments were performed 210 on three different days and the average percent recovery of the lipid standards is 211 provided in figure 5a and supplementary table 6.

212
To determine the coefficient of variation of all the lipid species, we extracted lipids from 213 plasma pooled from 5 individuals. For intra batch variations, the same sample was 214 subjected to mass spectrometric analysis 5 times. Lipidomics study in normal and vitamin B12 deficient human plasma 229 Vitamin B 12 , is a micronutrient mainly sourced from animal products, deficiency of which 230 has been reported to result in lipid imbalance 39 . Using this method, we attempted to 231 identify lipid species that are altered due to vitamin B12 deficiency. There was no    where Q1 is parent ion and Q3 is characteristic daughter ion, for that compound is better idea about the method, without any false detection hope (or lower detection 301 limits). In our method, DAG has highest LoD and LoQ of 1 nmol/L and 2nmol/L 302 respectively, which was still lower as compared to the previously reported methods for 303 targeted analysis 21 . The linearity of our method was found to be comparable to previous 304 lipidomics methods 21,27,41 .

305
The recovery of lipid species in our method was in the range of 69.75 % -113.19 %, 306 except DAG -137.5%, which were within the generally accepted range for quantification 307 and is comparable with other lipidomics studies 21,27 .

308
A major challenge in lipidomics experiments have been the high variability in the signals 309 and even the "shared reference materials harmonize lipidomics across MS-based 310 detection platforms and laboratories" have shown that most lipid species showed large 311 variability (CV) between 30% to 80% 44 . However variability for endogenous lipid species 312 that were normalized to the corresponding stable isotope-labelled analogue were lower 313 than 30% 40,44 . In this method, we used sum normalization (although we are not 314 addressing batch effect in this study) and found that 849 lipid species had a CV <30% 40 .

315
Overall, the median CV of our method (15.1%, 15.5%, and 14.7%), was similar to or 316 better than the previous reports 21,27,31 . In addition, we have also reported species-  To our knowledge, this is the first study to identify lipids with a significantly decreased

565
Lipid extraction from human plasma 566 We used a modified Bligh and Dyer method using Dichloromethane/methanol/water 567 (2:2:1 v/v). The study was approved by institutional ethical committee of CSIR-IGIB.

568
Human plasma (10 μL) was mixed with 490 μL of water (in glass tube) and incubated on 569 ice for 10 minutes. Lipid internal standard mixes (5 µL, consisting of splashmix and 570 ceramide) was added to a mixture of methanol (2 mL) and dichloromethane (1 mL); the 571 mixture was vortexed and allowed to incubate for 30 minutes at room temperature. After Liquid chromatography-Mass spectrometry: 582 We used an Exion LC system with a Waters AQUITY UPLC BEH HILIC XBridge Amide 583 column (3.5 µm, 4.6 x 150 mm) for chromatographic separation. The oven temperature 584 was set at 35°C and the auto sampler was set at 4°C. Lipids were separated using 585 buffer A (95% acetonitrile with 10mM ammonium acetate, pH-8.2) and buffer B (50% 586 acetonitrile with 10mM ammonium acetate, pH-8.2) with following gradient: with a flow 587 rate of 0.7 ml/minute, buffer B was increased from 0.1% to 6% in 6 minutes, increased 588 to 25% buffer B in next 4 minutes. In the next 1 minute buffer B was ramped up to 98%, 589 further increased to 100% in the next 2 minutes, and held at the same concentration 590 and flow rate for 30 seconds. Flow rate was increased from 0.7 ml/min to 1.5 ml/min 591 and 100% buffer B was maintained for the next 5.1 minutes. Buffer B was brought to The raw analytical signal obtained for standards from plasma lipid extract (spiked with 659 standards) was used to estimate the LoD and LoQ, using the following formula:

662
The standard solution was diluted serially with matrix and the lipid standards were run in  Figure 3.